Molecular Surveillance of European Quinolone-Resistant Clinical Isolates of Pseudomonas aeruginosa and Acinetobacter spp. Using Automated Ribotyping

Abstract

Nosocomial isolates of Pseudomonas aeruginosa and Acinetobacter spp. exhibit high rates of resistance to antibiotics and are often multidrug resistant. In a previous study (D. Milatovic, A. Fluit, S. Brisse, J. Verhoef, and F. J. Schmitz, Antimicrob. Agents Chemother. 44:1102–1107, 2000), isolates of these species that were resistant to sitafloxacin, a new advanced-generation fluoroquinolone with a high potency and a broad spectrum of antimicrobial activity, were found in high proportion in 23 European hospitals. Here, we investigate the clonal diversity of the 155 P. aeruginosa and 145 Acinetobacter spp. sitafloxacin-resistant isolates from that study by automated ribotyping. Numerous ribogroups (sets of isolates with indistinguishable ribotypes) were found among isolates of P. aeruginosa (n = 34) and Acinetobacter spp. (n = 16), but the majority of the isolates belonged to a limited number of major ribogroups. Sitafloxacin-resistant isolates (MICs > 2 mg/liter, used as a provisional breakpoint) showed increased concomitant resistance to piperacillin, piperacillin-tazobactam, ceftriaxone, ceftazidime, amikacin, gentamicin, and imipenem. The major ribogroups were repeatedly found in isolates from several European hospitals; these isolates showed higher levels of resistance to gentamicin and imipenem, and some of them appeared to correspond to previously described multidrug-resistant international clones of P. aeruginosa (serotype O:12) and Acinetobacter baumannii (clones I and II). Automated ribotyping, when used in combination with more discriminatory typing methods, may be a convenient library typing system for monitoring future epidemiological dynamics of geographically widespread multidrug-resistant bacterial clones.

Pseudomonas aeruginosa and Acinetobacter spp. are ubiquitous in natural environments and are common opportunistic pathogens in hospitals. In contrast to community-acquired strains, nosocomial strains exhibit high rates of resistance to antibiotics and are often multidrug resistant (3, 7, 37). Until the end of the 1980s, fluoroquinolones had excellent activities against P. aeruginosa and Acinetobacter spp., but extensive use of these antimicrobials, in particular ciprofloxacin, has led to an increasing incidence of ciprofloxacin-resistant isolates in many bacterial species, especially in P. aeruginosa, Acinetobacter spp., and Staphylococcus aureus (15, 22, 50). Sitafloxacin (DU-6859a) is a new advanced-generation fluoroquinolone with high potency and a broad spectrum of antimicrobial activity. However, in a previous study (32), isolates resistant to sitafloxacin were found in high proportion within some species, mostly in P. aeruginosa and Acinetobacter spp. Increasing incidence of these organisms might seriously limit therapeutic options since sitafloxacin-resistant isolates were always cross-resistant to all other quinolones tested (32) and since concomitant resistance of quinolone-resistant strains to other antibiotic classes is common (50).

Numerous hospital outbreaks of multiple-drug-resistant P. aeruginosa and Acinetobacter spp. have been investigated by molecular typing methods, and evidence of nosocomial transmission has been gathered repeatedly, in either endemic or epidemic situations (4, 19, 35, 47, 57). In addition, molecular epidemiology studies have shown that a limited number of clones of these species are repeatedly found in European hospitals. One clone of P. aeruginosa serotype O:12 is widely distributed across European countries (31, 41), and progression of new multiple-drug-resistant genotypes associated with the serotype O:11 is being observed (49). Similarly, with Acinetobacter spp., comparison of strains causing outbreaks in northwestern European countries has shown that most outbreaks are caused by a limited number of clones (18), and one amikacin-resistant clone is widespread in Spanish hospitals (56). Molecular typing has also shown that particular clones of P. aeruginosa can be highly prevalent in both environmental and clinical samples (42).

In spite of these advances, the identification of bacterial clones of major clinical and epidemiological importance remains difficult due to the lack of standardization of characterization schemes. Consequently, the geographic distribution of antibiotic-resistant clones of P. aeruginosa and Acinetobacter spp., as well as the dynamics of their distribution, is not well known. Monitoring the epidemiological dynamics of such clones is important for understanding the factors that determine their spread or their prolonged survival in hospitals.

Ribotyping has been used extensively to characterize bacterial clones of most clinically important bacterial species (9, 23) and has also been found to be valuable in hospital investigation and epidemiological tracking, alone or combined with other typing techniques, of Acinetobacter spp. (8, 18, 21, 34, 45, 53) and P. aeruginosa (6, 24, 27, 35). Ribotypes have been shown to be stable in vitro and during outbreaks (1, 35). Due to automation and standardization, the RiboPrinter microbial characterization system (14, 46) is suited for rapid and high-throughput typing of bacterial strains and for multicenter, longitudinal molecular surveillance of major clones. The approach has proven successful for the epidemiological study of strains belonging to species such as P. aeruginosa, Listeria monocytogenes, Klebsiella spp., and Burkholderia cepacia (2, 12, 13, 38, 58).

The purposes of the present study were (i) to evaluate the genetic diversity revealed by automated ribotyping of quinolone-resistant European clinical isolates of P. aeruginosa and Acinetobacter spp., (ii) to identify highly prevalent ribotypes of sitafloxacin-resistant isolates and investigate their distribution among European hospitals, (iii) to investigate whether sitafloxacin-resistant isolates show concomitant resistance to other classes of antimicrobial compounds and whether the major ribotypes show increased levels of resistance, and (iv) to initiate the construction of an electronic fingerprint database of major P. aeruginosa and Acinetobacter spp. international clones which will allow for the follow-up of their epidemiological dynamics in space and time.

MATERIALS AND METHODS

Bacterial isolates.

A total of 615 P. aeruginosa isolates and 400 Acinetobacter spp. isolates were collected between January 1997 and January 1999, as part of the European SENTRY international surveillance program, from 23 university hospitals in 14 European countries, and from two hospitals in South Africa. Only one isolate per patient was included. The SENTRY program is a longitudinal surveillance program designed to monitor the predominant pathogens and antimicrobial resistance patterns of nosocomial and community-acquired infections nationally and internationally (39). The monitored infections include bacteremia (objective A), community-acquired respiratory tract infections due to fastidious organisms (Streptococcus pneumoniae, Haemophilus influenzae, and Moraxella catarrhalis) (objective B), nosocomial pneumonia (objective C), wound infections (objective D), and urinary tract infections (objective E).

Isolates were referred to a regional reference laboratory located at the Eijkman-Winkler Institute for Clinical Microbiology, University Medical Centre of Utrecht, Utrecht, The Netherlands. Participating centers were instructed to refer only the first 20 consecutive blood isolates of any species per month and only 50 consecutive isolates from pneumonia, wound, and urinary tract infections. All isolates received were immediately stored at −70°C in Microbank storage tubes (Pro-Lab Diagnostics, Neston, United Kingdom) until studied further. The isolates were reidentified in our laboratory to the species level (P. aeruginosa) or to the genus level (Acinetobacter spp.) using either the Vitek or the API system (BioMérieux Benelux BV, s'Hertogenbosch, The Netherlands).

Participating hospitals.

The following countries and hospitals participated in the study: Austria (Krankenhaus der Elisabethinen, Linz), Belgium (Hôpital Erasme, Brussels), France (Hôpital St. Joseph, Paris; Hôpital de la Pitié-Salpêtrière, Paris; Hôpital Edouard Herriot, Lyon; Hôpital A. Calmette, Lille), Germany (Freiburg University Hospital, Freiburg; Düsseldorf University Hospital, Düsseldorf), Greece (National University of Athens, Athens), Italy (University Hospital of Genoa, Genoa; University Hospital of Rome, Rome), The Netherlands (University Medical Centre of Utrecht, Utrecht), Poland (Jagiellonian University Hospital, Cracow; Warsaw University Hospital, Warsaw), Portugal (University Hospital of Coimbra, Coimbra), Spain (University Hospital of Seville, Seville; Hospital Ramon y Cajal, Madrid; Hospital de Bellvitge, Barcelona), Switzerland (Centre Hospitalier Universitaire Vaudois, Lausanne), Turkey (University Hospitals in Ankara, Ankara; Istanbul University Hospital, Istanbul), the United Kingdom (St. Thomas's Hospital Medical School, London), Albania (Tirana University Hospital, Tirana), Israel (University Hospital of Tel Aviv, Tel Aviv), and South Africa (Cape Town University Hospital, Cape Town; Pretoria University Hospital, Pretoria).

Susceptibility testing.

MICs were determined using reference broth microdilution methods by following recommended guidelines of the National Committee for Clinical Laboratory Standards (33) with cation-adjusted Mueller-Hinton broth. The inoculum was adjusted to 5 × 10⁵ CFU/ml. Plates were read after 20 to 24 h of incubation at 35°C in ambient air. The following antibiotics were tested: sitafloxacin, ciprofloxacin, trovafloxacin, levofloxacin, clinafloxacin, gatifloxacin, moxifloxacin, ampicillin, piperacillin, piperacillin-tazobactam, ceftriaxone, ceftazidime, imipenem, gentamicin, amikacin, and cotrimoxazole.

Selection of isolates for automated ribotyping.

We analyzed all 301 sitafloxacin-resistant isolates (MIC > 1 mg/liter) except one (Table 1). In addition to the sitafloxacin-resistant isolates, sitafloxacin-susceptible isolates were included in the analysis in the following manner: from each center which had sent 20 or more isolates of a given species, 15% of these strains were randomly selected, whether they were sitafloxacin resistant or not. Among these, 62 sitafloxacin-susceptible isolates were present and analyzed (Table 1). Acinetobacter reference strains (18) of outbreak clone I (RUH875, RUH3247, RUH3238, and RUH3282) and of outbreak clone II (RUH3240, RUH134, RUH3422, and RUH3245) were included for comparison. In addition, reference strains of different DNA groups (DG) of Acinetobacter spp. were included: ATCC 23055^T (DG 1, A. calcoaceticus), ATCC 19606^T (DG 2, A. baumannii), ATCC 19004 (DG 3), ATCC 17906^T (DG 4, A. haemolyticus), ATCC 17908^T (DG 5, A. junii), ATCC 17979 (DG 6), ATCC 17909^T (DG 7, A. johnsonii), NCTC5866^T (DG 8, A. lwoffii), ATCC 9957 (DG 9), ATCC 17924 (DG 10), ATCC 11171 (DG 11), SEIP (DG 12, A. radioresistens), ATCC 17905 (DG 13 sensu Bouvet and Jeanjean [11]), ATCC 17903 (DG 13 sensu Tjernberg and Ursing [51]), strain Bouvet 382 (DG 14), strain Bouvet 240 (DG 15), ATCC 17988 (DG 16), and strain Bouvet 942 (DG 17).

TABLE 1.

Heterogeneity of populations of sitafloxacin-resistant and -susceptible isolates

Speciesa	Sitafloxacin-resistant isolates					Sitafloxacin-susceptible isolates					Total
	No. of strains typed	No. of distinct ribogroups	No. of major ribogroupsb	No. of centers	Simpson's index	No. of strains typed	No. of distinct ribogroups	No. of major ribogroups	No. of centers	Simpson's index	Ribogroups	Simpson's index
Pseudomonas aeruginosa (156/459)	155	34	10	22	0.91	48	32	3	17	0.98	52	0.93
Acinetobacter spp. (145/255)	145	16	7	18	0.82	14	11	1	5	0.93	21	0.83

^aIn parentheses are indicated the total numbers of sitafloxacin-resistant/sitafloxacin-susceptible isolates found in the susceptibility testing analysis. 

^bMajor ribogroups are defined as ribogroups comprising at least four isolates. 

Automated ribotyping.

Automated ribotyping was performed using the RiboPrinter microbial characterization system (Qualicon Europe Ltd., Warwick, United Kingdom) according to the manufacturer's instructions (14, 46). Briefly, a quantity corresponding to a few colonies was picked from fresh, individual Colombia agar plates with 5% sheep blood, suspended in sample buffer, and heat treated at 80°C for 20 min to inhibit endogenous DNA degradation enzymes. The samples could then either be kept at −20°C or processed directly. Lysis buffer was added to the samples before they were loaded into the RiboPrinter. Within the instrument, bacterial DNA was digested with a chosen restriction enzyme, and fragments were separated by agarose gel electrophoresis, transferred directly to nylon membranes, and hybridized with a chemiluminescent labeled probe containing the rRNA operon (rrnB) from Escherichia coli. The pattern of restriction fragments being hybridized and emitting chemiluminescent light was then converted to digital information by a charge-coupled-device camera and stored in the RiboPrinter database. Each sample datum was normalized by using an adjacent standard marker set. Similarity coefficients were calculated based upon both band position and relative intensity. Output patterns were merged into a single ribogroup by using an initial threshold similarity value of 0.93. Subsequently, the threshold value converged, as a function of the size of the ribogroup, until the minimal similarity value was 0.90. The restriction enzyme EcoRI was selected for analysis of Acinetobacter spp., and PvuII was selected for analysis of P. aeruginosa isolates, based on available evidence from the literature of the high discriminatory power of manual ribotyping (21) and after comparative analysis of different candidate enzymes. Quality control was performed by running a control batch of eight samples each week, and stability of the ribotype was always observed after the strains were passaged in vitro for several months. Ribogroups were automatically given a code composed of three numbers corresponding, in order, to the instrument number (always 210), the batch number of the first isolate run that fell into this ribogroup, and the position of this isolate in the batch (positions 1 to 8).

Serotyping.

P. aeruginosa strains were classified in one of 16 serogroups by agglutination on slides using four mixtures of commercially available sera (mixture A, O:1 plus O:3 plus O:4 plus O:6; mixture C, O:9 plus O:10 plus O:13 plus O:14; mixture E, O:2 plus O:5 plus O:12 plus O:15; and mixture F, O:7 plus O:8 plus O:11 plus O:12) and subsequently using 16 individual anti-O antisera (Sanofi Diagnostics Pasteur, Marnes-La-Coquette, France). Strains were grown on tryptic soy agar for 18 to 24 h at 37°C. Bacteria were suspended in a drop of each serum mixture on a slide and examined for the appearance of frank agglutination for several minutes.

Data analysis.

For cluster analysis, the patterns were exported into GelCompar version 4.0 (Applied Maths, Kortrijk, Belgium) using the GelConvert utility (Qualicon Europe Ltd.). Clustering was performed using the unweighted-pair group method with arithmetic averages (UPGMA) on the basis of the matrix of Pearson correlation coefficients of densitometric curves and using an optimization parameter of 1.2% (identical to the optimization parameter used in the RiboPrinter system). The discriminatory index of ribotyping was estimated using Simpson's index of diversity (29).

RESULTS

Clonal diversity of sitafloxacin-resistant and -susceptible isolates.

In a former study, we tested seven different quinolones for their in vitro activities against recent clinical bacterial isolates (32). The highest numbers of sitafloxacin-resistant isolates (defined as isolates with a MIC of >1 mg/liter, used as a provisional breakpoint, and with an intermediate level of resistance of 2 mg/liter) were found for P. aeruginosa, with 156 out of 615 isolates tested (25%) being sitafloxacin resistant, and for Acinetobacter spp. with 145 out of 400 isolates tested (36%) being sitafloxacin resistant. Sitafloxacin-resistant isolates were always cross-resistant to all other quinolones tested, namely, ciprofloxacin, clinafloxacin, gatifloxacin, levofloxacin, moxifloxacin, and trovafloxacin (32).

The sitafloxacin-resistant isolates of P. aeruginosa (n = 155) and Acinetobacter spp. (n = 145) and 62 sitafloxacin-susceptible isolates (see Materials and Methods) were analyzed by automated ribotyping. All isolates proved to be typeable. Reproducibility was found to be 100% after 50 randomly selected isolates of different bacterial species, including Acinetobacter spp. and P. aeruginosa, were run two times independently. The ribogroups created by the system were checked by eye for consistency. In a few instances, different ribogroups were created by the software, although the pattern differences between them did not seem to justify ribogroup distinction, for example, when it was due to intensity differences of the background. These ribogroups were then manually merged, as is allowed by the data management software.

Table 1 compares, for each species, the sitafloxacin-susceptible and sitafloxacin-resistant isolates. A high number of distinct ribotypes was found for P. aeruginosa (52 ribogroups, including 34 ribogroups among the sitafloxacin-resistant isolates), as well as for Acinetobacter spp. (21 ribogroups, including 16 ribogroups among the sitafloxacin-resistant isolates). Simpson's discriminatory index ranged from 0.82 to 0.98. In both species, the discrimination index was lower among sitafloxacin-resistant isolates than among sitafloxacin-susceptible isolates (Table 1).

Epidemiology of sitafloxacin-resistant isolates and geographic distribution of major ribogroups.

Figures 1 and 2 illustrate the ribotypes obtained from isolates of P. aeruginosa and Acinetobacter spp., respectively, with the number of isolates belonging to each ribogroup. With P. aeruginosa, the number of isolates per ribogroup ranged from 1 to 43, and 11 ribogroups (here defined as major ribogroups) comprised four or more isolates. With Acinetobacter spp., seven major ribogroups were found and there were up to 52 isolates for ribogroup 210-41-3 (Table 1).

FIG. 1.

Overview of the 52 distinct PvuII ribotypes obtained after ribotyping of 203 P. aeruginosa clinical isolates, including 155 sitafloxacin-resistant isolates. The number of isolates falling in each ribogroup and the number of centers where a ribogroup was collected are indicated only when the numbers were >1. The serotype(s) determined for one or more strains within each ribogroup is indicated. Several isolates were serotyped only for ribogroups 210-88-5, 210-87-3, and 210-87-2. For cluster analysis, UPGMA was used based on the matrix of Pearson correlation coefficients. In the dendrogram scale, correlation levels were converted to percent similarity levels. Four dominant ribogroups, with 15 to 43 isolates, were found, with the most prevalent one being 210-87-3, which corresponds to the widespread genotype serotype O:12 (31, 41). NT, nontypeable; nd, not determined.

FIG. 2.

Overview of the 21 distinct EcoRI ribotypes obtained after the ribotyping of 159 Acinetobacter spp. clinical isolates, including 145 sitafloxacin-resistant isolates. The number of isolates falling in each ribogroup is indicated only when the number was >1. The DG or taxonomic species of reference strains analyzed for comparison are given on the right of the ribogroup to which they belong. Ribotypes of outbreak clones I and II (18) are indicated. Strain ATCC 19606^T, the type strain of A. baumannii, fell in ribogroup 210-52-7. The asterisks indicate that the species or DG was deduced based on the correspondence of patterns with those published previously (21, 43). DG 13TU, DG 13 sensu Tjernberg and Ursing (51). For cluster analysis, UPGMA was used based on the matrix of Pearson correlation coefficients. In the dendrogram scale, correlation levels were converted to percent similarity levels.

The geographic distribution of the major ribogroups is shown in Table 2. With P. aeruginosa, most ribogroups were geographically widespread, even when we consider only the sitafloxacin-resistant isolates (distributed in three to eight different countries). Only one ribogroup (210-88-1, n = 7) was restricted to a single hospital (in France). There was an obvious correlation between the size of the ribogroup and its geographic spread (Table 2). A wide geographic distribution was also observed for most major ribogroups of Acinetobacter spp. (Table 2). However, ribogroups 210-61-2 (n = 7) and 210-64-6 (n = 25) were each restricted to single hospitals in Spain.

TABLE 2.

Geographic distributions of the major ribogroups and antibiotic resistance characteristics of the isolates of each clonal-transmission case

Species	Ribogroup	Total no. of isolates	Total no. of sitafloxacin- resistant isolates	Country (or countries) of origin of the sitafloxacin-resistant isolates (no. of isolates)	Country (or countries) of origin of the sitafloxacin-susceptible isolates (no. of isolates)	Country, center, with clonal-transmission casesa (n)	Antibiotics to which isolates belonging to clonal-transmission cases were resistantb	Period of clonal transmission
Pseudomonas aeruginosa	210-87-3	43	41	France (24), Greece (1), Italy (1), Poland (1), Portugal (6), Spain (2), Turkey (1), South Africa (5)	Italy (1), The Netherlands (1)	Portugal (6)	All	July 1997–May 1998
						France, hospital 1 (7)	All but CAZ	March 1997–May 1998
						France, hospital 2 (15)	All but CAZ, IMI, and AMIK	April 1997–May 1998
	210-88-2	28	27	France (2), Greece (4), Italy (12), Poland (4), Spain (2), Turkey (3)	The Netherlands (1)	Italy (8)	All	April 1997–August 1998
	210-99-4	18	14	France (5), Turkey (9)	Germany (1), Spain (1), Switzerland (1), United Kingdom (1)	Turkey (9)	All	March 1997–November 1997
	210-88-5	15	11	France (7), Poland (4)	France (1), Italy (1), Spain (2)	None
	210-87-4	8	8	France (7), Germany (1)	None	None
	210-87-2	8	4	Austria (1), Belgium (1), France (1), Spain (1)	France (1), Portugal (1), Spain (1), United Kingdom (1)	None
	210-88-1	7	6	France (6)	Greece (1)	None
	210-87-8	6	5	France (1), Greece (3), Poland (1)	Turkey (1)	None
	210-99-2	6	3	France (1), Turkey (1), South Africa (1)	Italy (1), Switzerland (2)	None
	210-87-7	5	5	France (2), Portugal (3)	None	None
	210-95-3	4	4	France (4)	None	None
Acinetobacter spp.	210-41-3	52	48	Belgium (2), France (1), Germany (1), Greece (1), Poland (1), Portugal (1), Spain (39), Turkey (2)	Greece (1), Portugal (1), Turkey (1), South Africa (1)	Spain, hospital 1 (16)	All	January 1997–November 1998
					None	Spain, hospital 2 (20)	All but IMI	April 1997–October 1998
	210-64-6	25	25	Spain (25)	None	Spain, hospital 1 (25)	All	April 1997–November 1998
	210-52-4	24	24	France (5), Italy (2), The Netherlands (2), Spain (15)	None	Spain, hospital 1 (11)	All but IMI	April 1997–October 1997
	210-60-2	17	17	Italy (14), Poland (1), Spain (1), South Africa (1)	None	Italy (14)	All but IMI	January 1997–November 1998
	210-52-7	10	10	France (4), Greece (4), Portugal (2)	None	None
	210-61-2	7	6	Spain (6)	Spain (1)	None
	210-110-3	5	4	Poland (1), Spain (3)	Spain (1)	None

^aCases with at least six (for P. aeruginosa) or 11 (for Acinetobacter spp.) sitafloxacin-resistant isolates collected from the same hospital and belonging to the same ribogroup. 

^bCAZ, ceftazidime; IMI, imipenem; AMIK, amikacin. 

The recurrent isolation of a given ribogroup in a single hospital was often observed (see Table 2, clonal-transmission cases). With P. aeruginosa for example, there were five instances of 6 or more (from 6 to 15) isolates of indistinguishable ribotype being found in the same hospital. Similarly, with Acinetobacter spp., there were five instances of 11 or more (from 11 to 25) isolates being collected from the same hospital and belonging to the same ribogroup. These findings represent circumstantial evidence of within-hospital spread of sitafloxacin-resistant isolates, and cases of infection with isolates of these types will hereafter be referred to as clonal-transmission cases. In order to verify that these recurrent ribotypes were not dominant also in the sitafloxacin-susceptible population of the same hospital, we further analyzed two cases. First, we typed 13 randomly selected isolates out of the 24 P. aeruginosa sitafloxacin-susceptible isolates of French hospital 2, where 15 out of the 22 collected resistant isolates belonged to ribogroup 210-87-3 (Table 2). Second, we typed 12 randomly selected isolates out of the 17 sitafloxacin-susceptible Acinetobacter isolates collected in Spanish hospital 2, in which 20 out of the 32 collected resistant isolates belonged to ribogroup 210-41-3 (Table 2). In both instances, no sitafloxacin-susceptible isolate showed the same ribotype as the corresponding dominant ribogroup of resistant isolates.

Comparison of the major ribogroups did not reveal significant differences in their clinical origins. All major ribogroups were isolated from blood samples, and most of them were also isolated from other clinical samples (see Materials and Methods), with frequencies being correlated to the size of the ribogroup.

Concomitant resistance of quinolone-resistant isolates and isolates of major ribogroups to other antibiotic classes.

We first investigated whether sitafloxacin-resistant isolates also showed concomitant resistance to other compounds, besides the well-known cross-resistance to all fluoroquinolones (32). With P. aeruginosa, 14% of sitafloxacin-susceptible isolates (n = 459) were resistant to piperacillin (MIC > 64 mg/liter) whereas this was the case for 69% of the sitafloxacin-resistant isolates (n = 156). Similarly, increases in percentages of resistant isolates were observed for piperacillin-tazobactam (10 versus 63%), ceftriaxone (95 versus 98%), ceftazidime (20 versus 61%), amikacin (4 versus 47%), gentamicin (16 versus 83%), and imipenem (11 versus 48%). These differences, with the exception of ceftriaxone, were highly significant as revealed by the chi-square test (P < 0.0001). Similarly, with Acinetobacter spp., 61% of sitafloxacin-susceptible isolates (n = 255) were resistant to piperacillin whereas this was the case for 100% of the sitafloxacin-resistant isolates (n = 145). Increases in percentages of resistant isolates were also observed for piperacillin-tazobactam (41 versus 90%), ceftriaxone (68 versus 99%), ceftazidime (48 versus 97%), amikacin (36 versus 86%), gentamicin (47 versus 97%), and imipenem (12 versus 46%). All differences were highly significant (P < 0.0001).

Second, we investigated whether the major ribogroups showed higher levels of concomitant resistance than those of the sporadic ribogroups (n < 4). Compared to the sitafloxacin-resistant isolates belonging to the sporadic ribogroups, the sitafloxacin-resistant isolates belonging to the major ribogroups showed, on average, a higher level of resistance to gentamicin and imipenem. With P. aeruginosa, only 7 out of 30 isolates (23%) belonging to sporadic ribogroups were resistant to imipenem whereas this was the case for 68 out of 125 isolates (54%) belonging to major ribogroups. Similarly, only 18 out of 30 isolates (60%) belonging to sporadic ribogroups were not susceptible to gentamicin (MIC > 8 mg/liter), whereas this was the case for 111 out of 125 isolates (89%) belonging to major ribogroups. With Acinetobacter spp., only 2 out of 12 isolates (16%) belonging to sporadic ribogroups were resistant to imipenem whereas this was the case for 64 out of 133 isolates (48%) belonging to major ribogroups. Finally, differences in levels of resistance to gentamicin were found between isolates of sporadic ribogroups (10 out of 12, 83%) and those of major ribogroups (131 out of 133, 98%). All comparisons were significant according to the chi-square test (P < 0.05). Within the major ribogroups, the degree of resistance observed in the clonal-transmission cases was not higher than that observed in the isolates not associated with clonal-transmission cases.

Antibiotic resistance patterns of the major clones within and among hospitals.

Although we found a few exceptions, the resistance profiles of all isolates belonging to a given case of clonal transmission were generally indistinguishable. For example, with Acinetobacter spp., all isolates belonging to ribogroup 210-60-2 from a single center in Italy were resistant to all antibiotics tested except imipenem (Table 2).

When a given ribogroup was implicated in clonal transmission in several hospitals, the susceptibility patterns of the isolates were homogeneous within hospitals but differed between hospitals. With P. aeruginosa, ribogroup 210-87-3 was associated with three clonal-transmission cases. Isolates coming from the hospital in Portugal were resistant to all antibiotics tested, while those isolates coming from one hospital in France were susceptible only to ceftazidime and those isolates coming from another hospital in France were susceptible to ceftazidime, imipenem, and amikacin. With Acinetobacter, ribogroup 210-41-3 was resistant to all antibiotics in one hospital but resistant to all except imipenem in another hospital.

Finally, in a single hospital in Spain, the three Acinetobacter ribogroups 210-41-3, 210-64-6, and 210-52-4 coexisted over a period of 6 months. All isolates of the two first ribogroups were resistant to all drugs tested, including imipenem, whereas isolates of the last ribogroup were resistant to all antimicrobial agents except imipenem.

Correspondence of the major ribogroups with previous typing results.

The possibility that the major sitafloxacin-resistant ribogroups identified herein corresponded to the previously identified major clones of clinical isolates of P. aeruginosa and Acinetobacter spp. was investigated. The serotype of one randomly selected isolate of each P. aeruginosa ribogroup was determined for most of the ribogroups and is indicated in Fig. 1. Eleven different serotypes were found (Fig. 1). The most common serotype found was O:11, distributed among 10 ribogroups, among which several corresponded to major ribogroups (Fig. 1). Serotype O:12 was found only in ribogroup 210-87-3. In order to check for the stability of the serotype within a given ribogroup, 17 randomly selected isolates of ribogroup 210-87-3 from seven countries, as well as 10 isolates of ribogroup 210-88-5 and the eight isolates of ribogroup 210-87-2, were serotyped. Isolates of ribogroup 210-87-3 were characterized as serotype O:12, with only two exceptions: one isolate was nontypeable, and one isolate showed serotype O:4. In contrast, isolates of ribogroup 210-87-2 were characterized as serotype O:1 (three isolates) and serotype O:6 (one isolate) and four isolates were nontypeable. In ribogroup 210-88-5, isolates were characterized as O:11 (n = 2), O:6 (n = 2), or O:1 (n = 1) and four isolates were nontypeable. Thus, the serotypes of isolates within a given ribogroup did not appear to be homogeneous, with the exception of that of ribogroup 210-87-3.

Comparison of A. baumannii strains causing outbreaks in northwestern European countries has indicated that most outbreaks are caused by two clones, named clone I and clone II (18). Ribotyping analysis of four isolates of each of these two clones was performed. All four isolates of clone I belonged to ribogroup 210-60-2, whereas all four isolates of clone II fell into ribogroup 210-41-3 (Fig. 2).

Gerner-Smidt (21) showed that ribotyping with EcoRI yielded banding patterns that were specific for strains in a given DG. Correspondence between some ribotypes presently obtained by automated ribotyping with those obtained by manual ribotyping (21, 45) suggested a presumptive identification for some isolates at the DG level (Fig. 2). In addition, cluster analysis of EcoRI ribotypes grouped Acinetobacter strains according to their DG (21). Here, the patterns of most ribogroups (14 out of 21, representing 152 out of 159 isolates) clustered into a single branch containing the A. baumannii type strain and representatives of A. baumannii clones I and II at a similarity level of more than 55% (Fig. 2). These results indicate that this cluster, in which all major ribogroups were included, may correspond to A. baumannii.

DISCUSSION

By applying automated ribotyping to a broad collection of sitafloxacin-resistant isolates, we identified for both species analyzed major groups of multidrug-resistant isolates that were collected from diverse geographic origins but showed indistinguishable ribotypes. These sets of isolates can be equated to clones, i.e., sets of isolates derived from a common ancestral cell, since it is very unlikely that similar ribotype patterns arise independently by convergent evolution (9).

Evidence of emergence of resistance during treatment with quinolones has been reported (36). Indeed, since decreasing levels of susceptibility to quinolones can result from only one or a few stepwise alterations in the target proteins of these drugs, namely, gyrase and topoisomerase IV (40, 54, 55), resistant mutants arise frequently enough to be selected in the course of treatment. However, once a resistant strain has emerged, it can also disseminate rapidly among patients, as was reported for P. aeruginosa (5, 15) and Acinetobacter spp. (28). Our data provide circumstantial evidence of numerous cases of within-hospital transmission of quinolone resistance, under endemic rather than epidemic situations, since resistant isolates with indistinguishable ribotypes were collected over periods of up to 23 months in single hospitals and since these dominant ribotypes were not encountered in the population of susceptible isolates from the same hospitals. It may still be that strains characterized by the same ribotype in fact show genetic differences when more discriminatory typing methods are used. However, with Acinetobacter, little additional variation was found among outbreak strains of the same ribotype but originating from different countries, using amplification of fragment length polymorphism (AFLP) analysis (18). This suggested that the two major clones, 210-41-3 (clone II) and 210-60-2 (clone I), are very homogeneous not only within hospitals but also among hospitals and countries. Clonal relationships of epidemiologically unrelated A. baumannii strains were also suggested on the basis of ribotyping (21) and repetitive extragenic palindromic PCR (56). Still, it is probable that quinolone resistance has emerged independently in several hospitals in isolates belonging to dominant and ubiquitous clones, and our results can therefore not be interpreted as evidence for clonal spread among hospitals of a single, initially sitafloxacin-resistant strain.

Isolates of a given ribogroup showed different antibiotic profiles in different hospitals, whereas different ribogroups shared the same antibiotic profile in a given hospital. Dijkshoorn et al. (18) also found different antimicrobial resistance profiles among outbreak strains from different hospitals belonging to the same clone. This finding indicates the importance of the hospital environment, probably antibiotic usage, rather than the strain, in determining the antimicrobial susceptibilities of outbreak strains, which can be rapidly modified by the acquisition or loss of plasmids. Since isolates belonging to the major ribogroups showed higher degrees of resistance to imipenem and gentamicin than those of isolates of the sporadic ribogroups, and since these two compounds are often used in infections by these organisms, antibiotic pressure is probably a determining factor favoring the selection of the outbreak strains.

Comparison of the within-hospital ribotype complexities and geographic distributions of the ribogroups showed contrasting situations for both species analyzed. With Acinetobacter, a limited number of ribotypes were found in most hospitals and ribotypes were sometimes specific for one or a few hospitals, whereas numerous P. aeruginosa ribotypes were collected with low frequency from most hospitals, and ribotypes appear geographically widespread. These results are consistent with a high rate of clonal transmission of Acinetobacter from other patients, from care personnel, or from the hospital environment and a frequent origin of P. aeruginosa infective strains directly from patients themselves (10, 47).

The results suggest correspondence between the major ribogroups of sitafloxacin-resistant isolates identified in the present study and the previously identified major clones of P. aeruginosa and A. baumannii. First, strains with ribotypes corresponding to those of clone I and clone II previously found in northwestern Europe (18) and in the Czech Republic (34) were found in our study, with strains of ribogroup 210-60-2 in Italy, Poland, Spain, and South Africa, and with strains of ribogroup 210-41-3 in almost all countries, including South Africa. This suggests a wide geographic spread of clones I and II, one that is possibly worldwide. Thus, our data are consistent with a limited number of A. baumannii clones being responsible for outbreaks in European hospitals (18). Second, the major ribogroup found in P. aeruginosa exhibited the serotype O:12 clone, in agreement with previous reports of the high prevalence of the clone harboring this serotype (31, 41). The present data also agree with previous reports that the O:11 serotype is associated with multiple epidemic strains and linked to various genetic backgrounds (49). However, follow-up of the correspondence of serogroups with P. aeruginosa ribogroups will be needed, as this relationship appears to be dynamic, possibly due to horizontal transfer of the genes which determine the serogroup (30, 43). Overall, these results show that automated ribotyping is useful for the identification of these major clones.

In a former study (32), we have shown that sitafloxacin, together with clinafloxacin, is one of the most active quinolones tested in vitro against both gram-positive and gram-negative clinical bacterial isolates collected in European university hospitals. The concomitant resistance of sitafloxacin-resistant isolates to other classes of antimicrobial compounds and the fact that these strains correspond to a limited number of major clones with high epidemic potential call for the molecular surveillance of these clones. Molecular surveillance will be required to monitor the spread and prevalence of the clones over extended periods of time, to understand better their epidemiological dynamics, and to identify clones with particular transmissibility or virulence potential. Surveillance programs make use of library typing systems, in which strains characterized at different times and locations can be reliably compared (48). Important facts about the epidemiological dynamics of major clones have been gathered using library typing systems for methicillin-resistant Staphylococcus aureus (17, 52), S. pneumoniae (20, 26), and Mycobacterium tuberculosis (16). For example, surveillance over long periods of time has demonstrated the replacement of certain major methicillin-resistant Staphylococcus aureus clones by other clones in Portugal (44).

Library typing systems must score evolutionary changes that are slow enough for large-scale surveillance yet discriminatory enough to track epidemic clones (48). With P. aeruginosa, the discriminatory power of ribotyping when it is performed with the enzyme PvuII has been claimed to be better than when it is performed with the enzyme EcoRI (27) and comparable to (6, 27) or slightly less than (24) that of pulsed-field gel electrophoresis (PFGE). In addition, PvuII has been suggested to be more stable and reproducible than EcoRI (35). With Acinetobacter spp., ribotyping with EcoRI and by arbitrarily primed PCR were found to be equally discriminatory (53) and slightly less discriminatory than PFGE (45) and AFLP analysis (18) but still useful for outbreak investigation (8) and identification of clonally related but epidemiologically unrelated strains (18, 21). Here, we found that automated ribotyping has a high discriminatory index, similar to that found previously (21, 35). Together with its excellent reproducibility, typeability rate, and high-volume capacity, this approach appears to be a convenient way to rapidly identify and compare the prevalent bacterial clones in distant geographic regions and time points. In addition, it generates easily manageable databases of standardized fingerprints that can be easily exchanged between laboratories. To investigate further the genetic differences between strains belonging to a unique ribogroup, PFGE, AFLP analysis, and PCR fingerprinting (25) are more discriminatory typing systems, but they are also more time-consuming and of lower throughput. They could be optimally used on a restricted number of isolates selected on the basis of their ribotype.