Introduction. Journal of Antimicrobial Chemotherapy (2000) 45,

Transcription

1 Journal of Antimicrobial Chemotherapy (2000) 45, JAC Activity and spectrum of 22 antimicrobial agents tested against urinary tract infection pathogens in hospitalized patients in Latin America: report from the second year of the SENTRY Antimicrobial Surveillance Program (1998) Ana C. Gales a,b *, Ronald N. Jones a, Kelley A. Gordon a, Hélio S. Sader b, Werner W. Wilke a, Mondell L. Beach a, Michael A. Pfaller a, Gary V. Doern a and the SENTRY Study Group (Latin America) a Medical Microbiology Division, 251 MRC, Department of Pathology, University of Iowa College of Medicine, Iowa City, IA 52242, USA; b Laboratório Especial de Microbiologia Clínica, Division of Infectious Diseases, Universidade Federal de São Paulo, São Paulo, Brazil The potency and spectrum of various antimicrobial agents tested against 434 bacterial isolates causing urinary tract infection (UTI) in hospitalized patients in Latin America were evaluated. The genotypes of the extended-spectrum -lactamase-producing and selected multi-resistant isolates were also evaluated by molecular typing techniques. Escherichia coli (60.4%) was the most common aetiological agent causing UTI, followed by Klebsiella spp. (11.2%) and Pseudomonas aeruginosa (8.3%). In contrast, Enterococcus spp. isolates caused only 2.3% of UTIs. Fewer than 50% of E. coli isolates were susceptible to broad-spectrum penicillins. The resistance rates to ciprofloxacin and the new quinolones were also high among these isolates. The molecular characterization of ciprofloxacin-resistant E. coli showed that most of them have a double mutation in the gyra gene associated with a single mutation in the parc gene. The Klebsiella pneumoniae isolates studied demonstrated high resistance rates to -lactam drugs, including broad-spectrum cephalosporins. The carbapenems were the compounds with the highest susceptibility rate among these isolates (100.0% susceptible) followed by cefepime (91.7% susceptible). Meropenem, imipenem and cefepime were also the most active drugs against Enterobacter spp. Among P. aeruginosa isolates, meropenem (MIC 50, 2 mg/l) was the most active compound, followed by imipenem (MIC 50, 4 mg/l), cefepime (MIC 50, 8 mg/l) and ceftazidime (MIC 50, 16 mg/l). The results presented in this report confirm that bacterial resistance continues to be a great problem in Latin American medical institutions. Introduction Urinary tract infection (UTI) is one of the most common infectious diseases diagnosed in outpatients. It is estimated that 3 5 million office visits occur yearly owing to lower UTI in the United States. 1 It also constitutes the most common nosocomial infection in many hospitals, and accounts for approximately 35% of all hospital-acquired infections reported to the National Nosocomial Infection Surveillance (NNIS) system. 2 A high proportion of hospitalacquired UTIs are associated with indwelling catheters. This device subverts several host defences to allow bacterial entry into the urinary tract. 3 Many catheter-associated bacteriurias are asymptomatic, and fewer than 5% of them will be complicated by bacteraemia. However, UTIs are the most frequent source of bacteraemia in hospitalized patients. 4,5 Platt and colleagues concluded that the acquisition of UTI during indwelling bladder catheterization was also associated with nearly a three-fold increase in patient mortality. 6 Nosocomially acquired UTIs also contribute additional length of hospital stay and healthcare costs to those expected from the underlying disease alone. In the community and hospital settings the aetiology of UTIs and the antimicrobial susceptibility of urinary pathogens have been changing over the years. 7,8 Factors such as the changing patient population, extensive use and misuse *Corresponding author. Tel: ; Fax: ; galesa@mail.medicine.uiowa.edu 2000 The British Society for Antimicrobial Chemotherapy 295

2 A. C. Gales et al. of antimicrobial agents, could all contribute to changes in the microbial profile of urinary tract isolates. 9 Most cases of UTI in the hospital setting are initially treated empirically based on the frequency of potential pathogens, local antimicrobial resistance rates and illness severity. The use of inappropriate empirical therapy was found to be a predictor of mortality in patients who had bacteraemia originating from a urinary tract source. 5 Consequently, the establishment of worldwide antimicrobial resistance surveillance systems seems to be an important step in detecting the emergence of resistance patterns, helping in the selection of the most efficacious empirical therapy at the local level, and supporting the implementation of preventive measures. The SENTRY Antimicrobial Surveillance Program has been monitoring antimicrobial resistance worldwide among selected types of infection since January The selection of the participant medical centres was based on the principle that they should be sentinel laboratories in their respective regions. The main objective of the present study was to evaluate the potency and spectrum of antimicrobial agents tested against UTI pathogens isolated in hospitalized patients in Latin America in the second year of the SENTRY Antimicrobial Surveillance Program (1998). We also determined the mechanism of resistance to quinolones among the ciprofloxacin-resistant Escherichia coli. Materials and methods Bacterial strains A total of 434 bacterial UTI isolates collected from Latin American patients were evaluated in this study. Only one isolate per patient was included. Ten Latin American laboratories participated in the study, eight were hospital based and independent laboratories served the two remaining medical centres. The laboratories were distributed throughout six countries (nine cities) including São Paulo, Rio de Janeiro and Florianópolis in Brazil; Buenos Aires and San Isidro in Argentina; Santiago (two sites) in Chile; Medellin in Colombia; Caracas in Venezuela; and Mexico City in Mexico. Nine of the 10 sites were identical to those participating in the 1997 study. In 1998 the Uruguayan centre was replaced by a Venezuelan centre. Each participant laboratory contributed approximately 50 strains, consecutively collected in the period from January to March The participant medical centres were directed by protocol to collect isolates from consecutively hospitalized patients with symptomatic UTI confirmed by urine culture (colony count 10 5 cfu/ml). A summary description of demographic data such as patient s age, gender, ward and hospitalization in intensive care unit was obtained. The mode of acquisition, community or nosocomial, was also noted. The isolates were identified to the species level by the participant centre and sent to the monitoring laboratory (University of Iowa, Iowa City, IA, USA) for identific a t i o n c o n firmation and reference antimicrobial susceptibility t e s t i n g. Susceptibility testing Antimicrobial susceptibility testing was performed and results interpreted using the reference broth microdilution method as described by the NCCLS. 11 Antimicrobial agents were obtained from their respective USA manufacturers for testing. The activities of ciprofloxacin, ofloxacin, gatifloxacin and trovafloxacin against the 23 ciprofloxacinresistant E. coli were also evaluated by Etest (AB Biodisk, Solna, Sweden). Quality control was performed using strains from the American Type Culture Collection (ATCC). Isolates of Klebsiella pneumoniae, Proteus mirabilis and E. coli with increased MICs ( 2 mg/l) of ceftazidime and/or ceftriaxone and/or aztreonam were considered as possible extended-spectrum -lactamase (ESBL)-producing phenotypes according to NCCLS criteria. 12,13 The ESBL phenotype was confirmed by additional tests using Etest strips (AB Biodisk), containing the -lactam substrate (aztreonam, cefotaxime, cefpodoxime, ceftazidime) with or without clavulanate at a fixed concentration of 2 mg/l. The variations of the MIC of the -lactam alone and of -lactam clavulanate combination were compared. A reduction of the -lactam MIC of more than 2 log 2 dilutions ( four-fold) in the presence of clavulanate indicated ESBL production. 14 Molecular methods The ESBL-producing isolates and selected multi-resistant isolates were genotyped by ribotyping. An isolate was considered multi-resistant if it was resistant to at least five of the following antimicrobial agents: imipenem (MIC 8 mg/l), ceftazidime (MIC 32 mg/l), cefepime (MIC 32 mg/l), ciprofloxacin (MIC 4 mg/l), penicillin -lactamase inhibitor combinations (MIC 128 mg/l), tetracycline (MIC 8 mg/l) and trimethoprim sulphamethoxazole (MIC 2 mg/l). Ribotyping was performed using the RiboPrinter Microbial Characterization System (E. I. dupont de Nemours, Wilmington, DE, USA) according to the manufacturer s instructions. The patterns were electronically imaged, stored and compared. Pattern comparison was based on both position and signal intensity. Isolates with coefficients of similarity 0.9 were considered to have the same ribotype profile. 15 The strains with identical ribotype profiles were also typed using pulsed-field gel electrophoresis (PFGE). PFGE was performed using the restriction endonucleases SpeI and SmaI for Enterobacteriaceae and Acinetobacter spp. isolates, respectively, as previously described. 16 Analysis of PFGE patterns was performed by visual inspection of photographs of ethidium bromide-stained gels. The isolates were classified as identical if they had the same bands, and as similar if they differed by one to three bands. Similar patterns were grouped under the same DNA type 296

3 Pathogens causing UTI in Latin America (subtypes) and isolates differing by more than three bands were considered to represent distinct DNA types. Twenty-three isolates were selected from 58 ciprofloxacin-resistant E. coli (MICs 8 mg/l) for evaluation of the mechanisms of resistance to quinolones. Each centre contributed at least one isolate, except for centres 39 and 49. The molecular characterization of quinolone resistance was determinated by amplification of the gyra and parc genes by PCR followed by sequencing of the respective amplicons. Briefly, PCR amplification of the genes was performed as follows: stock quantities of master mixes were prepared robotically. For the parc gene, 50 pmol/l of oligonucleotide primers, sense 5 -TCTGAACTGGGC- CTGAATG-3 (19-mer) and antisense 5 -CGTTCACCA- GCAGGTTAG-3 (18-mer), amplified a 344 bp fragment. For the gyra gene the primers, sense 5 -CCGTCGCG- TACTTTACGC-3 (18-mer) and antisense 5 -CGTTCA- CCAGCAGGTTAG-3 (18-mer), amplified a 384 bp fragment. 17 The PCR mixtures contained 200 mmol datp, dttp, dgtp and dctp; 50 mm NH 4 Cl; 1.5 mm MgCl 2 and 10 mm Tris HCl buffer ph 9.0 M at room temperature. 18 Aliquots of 45 L were robotically loaded into a 96-well polycarbonate plate (Corning, Corning, NY, USA), which was kept at 4 C in a mini-refrigerator. Evaporation and contamination were avoided by utilizing 35 L light mineral oil as an overlay above the 15 L of target template DNA. 19 Taq polymerase (2.5 U) (Promega, Madison, WI) were added to the master mix immediately before the assembling procedure by the robotic system. Thermal cycling conditions were as follows: 10 min at 95 C (initial DNA denaturing step); 1 min at 94 C (DNA denaturing), 1 min at 58 C (annealing), 3 min at 72 C (extension) cycling 35 times. PCR products were analysed on an agarose gel. Detection was accomplished by staining the products with ethidium bromide. Purification of the PCR products for cycle sequencing (Sanger technique) was accomplished with a QIAquick PCR purification kit (QIA- GEN, Valencia, CA, USA). Quantification of the purified DNA was performed with a Hitachi U2000 spectrophotometer, which was interfaced with the XP robotic system to avoid manual contact with amplified DNA. Additional electrophoretic separation ensured proper purification followed by adequate DNA dilution and mixing with the s e n s e primer for the sequencing reaction with subsequent detection on an ABI sequencer. Results and discussion The frequency of pathogens causing UTIs in Latin America is listed in Table I. The distribution of species was as follows: 262 E. coli (60.4%), 48 K. pneumoniae (11.1%), 36 Pseudomonas aeruginosa(8.3%), 20 P. mirabilis(4.6%), 14 Enterobacter spp. (3.2%), 14 Enterococcus spp. (3.2%), 10 Acinetobacter spp. (2.3%), 8 indole-positive Proteae spp. (1.8%), 7 Citrobacter spp. (1.6%) and 6 Serratiamarcescens (1.4%). The distribution of the five most frequent pathogens was very similar among the distinct Latin American centres. The Enterobacteriaceae were the most frequent pathogens detected, causing 84.3% of the UTIs. Previous studies have also demonstrated that E. coli is the most Table I. Frequency of pathogens causing urinary tract infections in hospitalized patients in Latin America (SENTRY Antimicrobial Surveillance Program, 1997 and 1998) No. of isolates (%) Rank order a Organism E. coli 262 (60.4) 246 (52.5) 2 Klebsiella spp. b 48 (11.1) 60 (12.8) 3 P. aeruginosa 36 (8.3) 35 (7.5) 4 P. mirabilis 20 (4.6) 23 (4.9) 5 Enterobacter spp. 14 (3.2) 20 (4.3) 6 Enterococcus spp. 14 (3.2) 20 (4.3) 7 Acinetobacter spp. 10 (2.3) c 15 (3.2) 8 Indole-positive Proteae 8 (1.8) d 11 (2.3) 9 Citrobacter spp. 7 (1.6) e 12 (2.6) 10 S. marcescens 6 (1.4) 9 (1.9) 11 other species 9 (2.1) f 18 (3.8) a Based on the 1998 statistics. b Only K. pneumoniae strains were isolated in c Includes nine A. baumannii strains. d Includes Morganella morganii (two strains), Proteus penneri (two strains), Proteus vulgaris (two strains) and Providencia stuartii (two strains). e Includes Citrobacter freundii (three strains) and Citrobacter koseri (four strains). f Includes Burkholderia cepacia (one strain), Pseudomonas fluorescens (one strains), Salmonella spp. (one strain), coagulase-negative staphylococci (two strains) and Staphylococcus aureus (four strains). 297

4 A. C. Gales et al. frequent aetiological agent causing community- and hospital-acquired UTIs. 2,9,20 22 It is interesting to observe that Enterococcus spp. isolates (3.2%) do not constitute an important cause of UTI in Latin America in contrast to results observed in North America. 22 The demographic data showed that 47.5% of the total number of pathogens causing UTI were isolated from patients with community-acquired infections, while 22.1% were from patients with nosocomially acquired infection. Unfortunately we were not able to classify the mode of acquisition of 30.4% of UTIs. Although urinary samples of children were also included in the study, the great majority of pathogens were isolated from adult patients (98.0%), principally women (72.0%). It has been extensively reported that adult women have a higher prevalence of UTI than men, principally owing to anatomic and physical factors. 23 The antimicrobial potency and spectrum for 22 selected antimicrobial agents against the five most frequent UTI pathogens are summarized in Table II. For E. coli isolates, the best coverage was achieved with carbapenems (100.0%), followed by amikacin (98.1%) and extended-spectrum cephalosporins ( %). Piperacillin tazobactam showed the highest activity (MIC 50, 4 mg/l) and susceptibility rate (88.5%) among the penicillin -lactamase inhibitor combinations. In spite of the high potency of the fluoroquinolones against E. coli isolates, a high resistance rate to ciprofloxacin was observed. Our study also made evident the existence of cross-resistance between ciprofloxacin and the newest quinolones such as levofloxacin and gatifloxacin. -Lactam resistance rates were higher among the 48 K. pneumoniae isolates evaluated. Only 68.8% of these isolates were susceptible to piperacillin tazobactam (MIC 50, 4 mg/l). The third-generation cephalosporins, ceftriaxone and ceftazidime, showed high potencies (MIC 50 s, 0.25 mg/l and 0.25 mg/l), but demonstrated elevated resistance rates (14.6% and 20.8%, respectively). Among the - lactams, only cefepime and the carbapenems exhibited susceptibility superior to 90%. The production of ESBL among E. coli and K. pneumoniae contributed significantly to the resistance of these isolates to the third- and fourthgeneration cephalosporins. However, other mechanisms of resistance such as plasmid-mediated ampc -lactamases and/or reduced outer membrane permeability could be involved in the resistance to -lactams since approximately 14.0% of the E. coli and 13.0% K. pneumoniae isolates were also resistant to cefoxitin. 24,25,26 Against P. aeruginosa isolates, although meropenem (MIC 50, 2 mg/l), imipenem (MIC 50, 4 mg/l) and cefepime (MIC 50, 8 mg/l) have been the most potent compounds, the highest percentage of susceptibility was found for piperacillin tazobactam (80.6%) followed by meropenem (77.8%). The low susceptibility rate (52.8%) displayed by imipenem might be due to dissemination of multi-resistant clones in the medical centres evaluated. The production of -lactamases coupled with loss of carbapenem-specific porins (OprD) could be the mechanisms involved in carbapenem resistance. Generally, severe P. aeruginosa infections are treated empirically with an association of a -lactam with an aminoglycoside; however, the low susceptibility rates to aminoglycosides could limit the effectiveness of such combined therapy in Latin America. Among the P. mirabilis isolates evaluated, low susceptibility (70.0%) to fluoroquinolones was also observed. Aztreonam, cefoxitin, ceftazidime and imipenem were active against all P. mirabilis isolates. Enterobacter spp. were resistant to the majority of -lactam drugs, and although cefepime showed higher activity than the other cephalosporins tested, only the carbapenems were able to inhibit the growth of 100.0% of Enterobacter spp. isolates. The overall antimicrobial susceptibility of the UTI pathogens isolated in Latin America was meropenem (98.1%) imipenem (95.9%) cefepime (89.6%) amikacin (87.7%) ceftazidime (85.7%) aztreonam (83.8%) ceftriaxone (81.6%) piperacillin tazobactam (77.5%) tobramycin (75.1%) and gentamicin (74.9%). The fluoroquinolone susceptibility rates ranged from 71.3% (grepafloxacin, data not shown) to 74.2% (gatifloxacin). The high resistance rates to fluoroquinolones among UTI pathogens in Latin America is of great concern, since these antimicrobial agents are agents frequently used in the treatment of UTIs. Of 34 enteric bacilli (18 E. coli, 12 K. pneumoniae and four P. mirabilis isolates) identified as ESBL phenotypes using the NCCLS criteria, 11 only 20 were characterized as possible ESBL producers by the Etest strip (Table IV). The performance of various -lactams in detecting ESBL-producing isolates was evaluated, since different types of ESBL can have different preferred -lactam substrates. 12 Cefotaxime exhibited the highest sensitivity (90.0%) and detected the greatest number of ESBL producers among the isolates evaluated, while aztreonam exhibited the lowest sensitivity rate (65.0%). Although ceftazidime (85.0%) displayed a lower sensitivity rate than cefotaxime, it was able to detect two isolates that were not detected by any other -lactam. It has been reported that ceftazidime is the preferred substrate for most ESBLs isolated in the USA and Europe. 12,27 However, our results agree with previous Latin American studies that reported cefotaxime as the preferred substrate in Latin America. 28,29 Based on the degradation of preferred substrates, tests for detection of ESBL-producing isolates have been developed. This finding suggests that the use of both substrates (ceftazidime and ceftriaxone) would allow the detection of most ESBLproducing isolates emerging in Latin America. Table III shows the co-resistance phenotypes among the isolates characterized as ESBL producers. Most of the ESBL-producing strains were resistant to an average of three different classes of antimicrobial agent. Co-resistance to tetracycline was the most frequent followed by trimethoprim sulphamethoxazole, tobramycin and flu o r o q u i n o l o n e 298

5 Pathogens causing UTI in Latin America 299

6 A. C. Gales et al. Table III. Use of the Etest to categorize and type suspected ESBL-producing enteric bacilli a MIC/MIC with 2 mg/l clavulanate (mg/l) Organism aztreonam cefotaxime cefpodoxime ceftazidime co-resistances b E. coli 211E 16/ / /1 32/ E 16/0.5 16/ /2 32/0.38 Cx,F,Tc,T/S 224E 16/0.5 16/ /4 32/0.75 Cx,F,Tc,T/S 300E 16/ / /1 16/0.5 Tc,To 348E c 16/ 0.5 3/ /4 32/0.25 F,Tc,To,T/S 414E c 16/ / 1 32/ 4 32/1.5 Cx,F,Tc,To,T/S 420E c 16/ / / 4 8/0.38 F,Tc K. pneumoniae 36E 16/0.5 16/ / /0.19 Tc,To,T/S 88E 0.25/ / /0.19 6/0.19 Tc,To,T/S 150E 16/ / / /0.25 Tc 167E 16/0.19 2/ /0.5 32/0.5 To 204E 16/ / / /1 Tc,To,T/S 263E 4/ / /0.25 1/0.25 F,To,T/S 276E 16/0.5 3/ / /0.75 Cx,F,Tc,To,T/S 281E 16/ / / /0.75 To,T/S P. mirabilis 69E 0.25/ / / /0.094 F,Tc,To,T/S 274E 0.25/ / /0.25 8/0.094 F,Tc,To,T/S 275E 0.25/ / /0.19 1/ F,Tc,To,T/S 325E 0.5/ / / /0.19 F,Tc,To,T/S a Nineteen confirmed ESBL strains were available for molecular and phenotypic typing. b Resistance was defined by NCCLS criteria: Cx, cefoxitin (MIC 32 mg/l); I, imipenem ( 16 mg/l); F, fluoroquinolones (ciprofloxacin MIC 4 mg/l); Tc, tetracycline (MIC 16 mg/l); To, tobramycin (MIC 16 mg/l) and T/S, trimethoprim sulphamethoxazole (MIC 4/76 mg/l). c The isolate could not be categorized as an ESBL producer because the MICs were out of scale. resistance. The antimicrobial resistance patterns observed among the ESBL-producing isolates confirm that the therapeutic options for treatment of infections caused by ESBL pathogens are limited to carbapenems. 30 Molecular epidemiological evaluation (ribotyping) was performed for 12 multi-resistant isolates (nine Acinetobacter baumannii and three Enterobacter cloacae) and 34 possible ESBL producers (18 E. coli, 12 K. pneumoniae and four P. mirabilis). The isolates with identical ribotypes are shown in Table IV. Among A. baumannii isolates, two identical ribotypes (521-1 and 815-2) were observed among six isolates from four different medical centres. Interestingly, isolates from different centres located in the same country (39/40 and 46/48) exhibited identical ribotypes and PFGE patterns, indicating possible inter-hospital dissemination of multi-resistant A. baumannii isolates. Inter-hospital transmission of A. baumannii has been described previously in Brazil. 31 Three multi-resistant E. cloacae originating from two medical centres were also evaluated. The E. cloacae isolates from the same centre shared identical ribotype and PFGE profiles, while the other isolate from a different centre showed a distinctly different ribotype. Great genomic variability was observed among the potential ESBL producers. As discussed previously, a high percentage of fluoroquinolone-resistant E. coli was observed in this surveillance study. Ciprofloxacin-resistant E. coli isolates were reported by all Latin American medical centres, except centres 39 and 49. The high ciprofloxacin resistance rate observed could be attributed in part to a higher prevalence of ciprofloxacin-resistant E. coli detected in one medical centre (25 of 58 ciprofloxacin-resistant E. coli isolates were obtained from medical centre 45). The resistance rates to fluoroquinolones vary from one country to another, and depend on local epidemiological factors. Generally, in the United States and Canada, E. coli isolates from patients with UTI display 95% susceptibility to fluoroquinolones. 22,23,32 However, a dramatic increase in the prevalence of fluoroquinolone resistance has been reported recently, in both community and nosocomial settings in several European countries In the present study, approximately 70.0% of the E. coli resistant to quinolones were collected from community-acquired UTIs. It could reflect the overuse of the quinolones for treatment of community- 300

7 Pathogens causing UTI in Latin America Table IV. Molecular epidemiological studies (automated ribotyping, PFGE) for multi-resistant bacteria from five species groups isolated at Latin American medical centres Molecular tests Medical centre Organism /isolate No. ribotype PFGE Acinetobacter spp. 39/ ACB-1 40/ ACB-1 40/ ACB-2 46/ ACB-3 48/ ACB-3 48/ ACB-3 E. cloacae 41/ ECL-1 41/ ECL-1 E. coli 45/ EC-1 46/ EC-2 K. pneumoniae 42/ KP-1a 42/ KP-1b 42/ KP-1c 48/ KP-2 P. mirabilis 39/ PM-1 40/ PM-2 42/ PM-3 42/ PM-3 acquired UTI. Some authors have advocated that quinolone resistance is higher in developing countries than in developed nations because of the use of less active quinolones, such as nalidixic acid, and/or the use of low dosages of more potent compounds such as ciprofloxacin resulting in selection of mutant isolates. 35 The use of short-term treatment with quinolones for UTI, which has been encouraged by some authors, could also have been a contributory factor for selection of mutant isolates. 36 On the other hand, other authors have recommended that the use of quinolones should be reduced or at least rationalized, principally among the UTIs, in order to save this potent class of antibiotics and avoid the development of resistance among the Enterobacteriaceae. 32 Mutations in gyra have commonly been localized near the amino terminus of the encoded protein, at positions 83 and ,38 The vast majority of E. coli isolates evaluated in this study (Table V) demonstrated double mutations in gyra at positions 83 (leucine replacing serine) and 87 (asparagine or tyrosine replacing aspartic acid). Among E. coli isolates, mutations in the parc gene contribute to quinolone resistance, but only in the presence of gyra mutations. All evaluated isolates also demonstrated mutations in the parc gene at positions 80 or 84. At position 80, isoleucine or arginine replaced the amino acid serine, while in position 84 glutamic acid was replaced by lysine. Different point mutations were detected in the isolates from centre 41, probably because this laboratory receives isolates from three different hospitals. Etest detected highlevel resistance to ciprofloxacin (MIC 32 mg/l) in all E. coli isolates with double mutations in gyra and a single mutation in parc. In general, these isolates also displayed high MICs for the most recently tested fluoroquinolones. However, the isolate that exhibited a single mutation in the gyra and parc genes demonstrated low-level resistance to ciprofloxacin (MIC, 6 mg/l) and trovafloxacin (MIC, 4 mg/l), but remained susceptible to gatifloxacin (MIC, 1.5 mg/l). These results indicate that multi-step mutations are predominantly responsible for the emergence of highlevel resistance to fluoroquinolones in the Latin American Centres monitored. For this reason, recognition of strains with low-level resistance is important. In this manner, the selection of high-level resistant isolates might be avoided and the activity of the newest quinolones might be preserved. In conclusion, the results presented in this report confirm that bacterial resistance continues to be a great problem in Latin American medical centres. We postulate that the results obtained in this study could be generalized for other Latin American medical centres that have the same epidemiological characteristics as ours. However, distinct local conditions such as local population, antimicrobial use Table V. The occurrences of topoisomerase gene mutations observed in 23 E. coli UTI strains with elevated fluoroquinolone MIC results (SENTRY Antimicrobial Surveillance Program, 1998) Coding changes MIC (mg/l) ranges by Etest parc gyra Medical centre number (occurrences) ciprofloxacin ofloxacin gatifloxacin trovafloxacin Ser80 Ile Ser83 Leu 41 (1) Ser83 Leu, Asp87 Asn 41 (1), 42 (1), 43 (3), 44 (2), 45 (1), 46 (3), (4) Ser83 Leu, Asp87 Tyr 41 (1), 44 (1) Ser80 Arg Ser83 Leu, Asp87 Asn 41 (1), 43 (1) 48 (1) Glu84 Lys Ser83 Leu, Asp87 Asn 40 (1), 41 (1)

8 A. C. Gales et al. and local infection control policies could result in differences in the aetiology and susceptibility profile of bacterial pathogens. The main problems detected in our study are carbapenem resistance among P. aeruginosa and Acinetobacter spp. isolates, ESBL-producing E. coli and Klebsiella spp., and rapidly increasing resistance to fluoroquinolones among Enterobacteriaceae, especially E. coli. Consequently, the following should be encouraged in Latin American hospitals and communities: (i) the optimization of local clinical microbiological laboratories; (ii) the rational use of antimicrobial drugs in the community, hospital and veterinary settings; and (iii) support for antimicrobial surveillance programmes at local and national levels. In this context, continuous programmes of surveillance such as SENTRY make possible continuous generation and analysis of comparative data complementing local programmes. Acknowledgements Kay Meyer provided excellent support in the preparation of the paper. We express our appreciation to all the medical technicians who have worked in SENTRY, especially to S. Coffman, M. Erwin, K. C. Kugler and A. O. Reis. The SENTRY Antimicrobial Surveillance Program has been sponsored by a research/educational grant from Bristol- Myers Squibb. Ana Cristina Gales is supported partially by a grant from CAPES, Brasília, Brazil. SENTRY Latin America Study Group includes: H. S. Sader (Brazil Latin America Coordinator); J. Sampaio (Laboratório Lâmina, Rio de Janeiro, Brazil); C. Zoccoli (Laboratório Médico Santa Lúzia, Florianópolis, Brazil); J. M. Casellas (Centro de Estudios en Antimicrobianos, San Isidro, Argentina); J. Smayevsky (Microbiology Laboratory CEMIC, Buenos Aires, Argentina); V. Prado (Faculdad de Medicina de Chile, Santiago, Chile); E. Palavecino (Universidad Catolica del Chile, Santiago, Chile); J. A. Robledo (Corporation Para Investigaciones Biologicas, Medellin, Colombia); J. Sifuentes-Osornio (Instituto Nacional de la Nutricion, Ciudad del Mexico, Mexico); and M. Guzman-Blanco (Caracas, Venezuela). References 1. Johnson, J. R. & Stamm, W. E. (1989). Urinary tract infections in women: diagnosis and treatment. Annals of Internal Medicine 111, National Nosocomial Infections Surveillance (NNIS) report, data summary from October 1986 April Issued May (1996). American Journal of Infection Control24, Warren, J. W. (1997). Catheter-associated urinary tract infections. Infectious Disease Clinics of North America 11, Krieger, J. N., Kaiser, D. L. & Wenzel, R. P. (1983). Urinary tract etiology of bloodstream infections in hospitalized patients. Journal of Infectious Diseases 148, Bishara, J., Leibovici, L., Huminer, D., Drucker, M., Samra, Z., Konisberger, H. et al. (1997). Five-year prospective study of bacteraemic urinary tract infection in a single institution. European Journal of Clinical Microbiology and Infectious Diseases 16, Platt, R., Polk, B. F., Murdock, B. & Rosner, B. (1982). Mortality associated with nosocomial urinary-tract infection. New England Journal of Medicine 307, Neu, H. C. (1992). Urinary tract infections. American Journal of Medicine 92, Suppl. 4A, S Jones, R. N. (1996). Impact of changing pathogens and antimicrobial susceptibility patterns in the treatment of serious infections in hospitalized patients. American Journal of Medicine 100, Suppl. 6A, S Brosnema, D. A., Adams, J. R., Pallares, R. & Wenzel, R. P. (1993). Secular trends in rates and etiology of nosocomial urinary tract infections at a university hospital. Journal of Urology 150, Pfaller, M. A., Jones, R. N., Doern, G. V. & Kugler, K. (1998). Bacterial pathogens isolated from patients with bloodstream infection: frequencies of occurrence and antimicrobial susceptibility patterns from the SENTRY antimicrobial surveillance program (United States and Canada, 1997). Antimicrobial Agents and Chemotherapy 42, National Committee for Clinical Laboratory Standards. (1997). Methods for Dilution Antimicrobial Susceptibility Tests for Bacteria That Grow Aerobically Fourth Edition: Approved Standard M7-A4. NCCLS, Wayne, PA. 12. Bush, K., Jacoby, G. A. & Medeiros, A. A. (1995). A functional classification scheme for -lactamases and its correlation with molecular structure. Antimicrobial Agents and Chemotherapy 39, National Committee for Clinical Laboratory Standards. (1998). Performance Standards for Antimicrobial Susceptibility Testing Eighth Informational Supplement: Approved Standard M100-S8. NCCLS, Wayne, PA. 14. Cormican, M. G., Marshall, S. A. & Jones, R. N. (1996). Detection of extended-spectrum -lactamase (ESBL)-producing strains by Etest ESBL screen. Journal of Clinical Microbiology 34, Pfaller, M. A., Wendt, C. & Hollis, R. J. (1996). Comparative evaluation of an automated ribotyping system versus pulsed-field gel electrophoresis for epidemiological typing of clinical isolates of Escherichia coli and Pseudomonas aeruginosa from patients with recurrent gram-negative bacteremia. Diagnostic Microbiology and Infectious Disease 25, Pfaller, M. A., Hollis, R. J. & Sader, H. S. (1992). Chromosomal restriction fragment analysis by pulsed-field gel electrophoresis. In Clinical Microbiology Procedures HandbookVol. 2. (Isenberg, H. D., Ed.), pp. 10.5c American Society for Microbiology, Washington, DC. 17. Georgiou, M., Muñoz, L. R., Román, F., Cantón, R., Gómez-Lus, R., Campos, J. et al. (1996). Ciprofloxacin-resistant Haemophilus influenzae strains possess mutations in analogous positions of GyrA and ParC. Antimicrobial Agents and Chemotherapy 40, Blanchard, M. M., Taillon-Miller, P., Nowotny, P. & Nowotny, V. (1993). PCR buffer optimization with uniform temperature regimen to facilitate automation. PCR Method and Applications 2, Wilke, W. W., Sutton, L. D. & Jones, R. N. (1995). Automation of polymerase chain reaction tests to achieve acceptable contamination rates. Clinical Chemistry 41,

9 Pathogens causing UTI in Latin America 20. Weber, G., Riesenberg, K., Schlaeffer, F., Peled, N., Borer, A. & Yagupsky, P. (1997). Changing trends in frequency and antimicrobial resistance of urinary pathogens in outpatient clinics and a hospital in Southern Israel, European Journal of Clinical Microbiology and Infectious Diseases 16, Sader, H. S., Jones, R. N., Winokur, P. L., Pfaller, M. A., Doern, G. V., Barrett, T. & The SENTRY Study Group, Latin America. (1999) Antimicrobial susceptibility of bacteria causing urinary tract infections in Latin American hospitals: Results from the SENTRY antimicrobial surveillance program (1997). Clinical Microbiology and Infection 5, Sader, H. S., Gales, A. C., Schomberg, L., Jones, R. N. & SEN- TRY Participants Group. (1998). Comparative evaluation of frequencies and antimicrobial susceptibilities of pathogens isolated from urinary tract infection (UTI) in North America (NA) and Latin America (LA): Results of the SENTRY Antimicrobial Surveillance Program, In Program and Abstracts of the Thirty-Eighth Interscience Conference on Antimicrobial Agents and Chemotherapy, San Diego, CA, Abstract E-39, p. 44. American Society for Microbiology, Washington, DC. 23. Barnett, B. J. & Stephens, D. S. (1997). Urinary tract infection: an overview. American Journal of Medical Science 314, Martinez-Martinez, L., Hernandez-Alles, S., Alberti, S., Tomas, J. M., Benedi, V. J. & Jacoby, G. A. (1996). In vivo selection of porindeficient mutants of K. pneumoniae with increased resistance to cefoxitin and extended-spectrum cephalosporins. Antimicrobial Agents and Chemotherapy 40, Pangon, B., Bizet, C., Bure, A., Pichon, F., Phillipon, A., Regnier, B. et al. (1989). In vivo selection of a cephamycin-resistant, porin-deficient mutant of Klebsiella pneumoniae producing a TEM-3 -lactamase. Journal of Infectious Diseases 159, Jenks, P. J., Hu, Y. M., Danel, F., Mehtar, S. & Livermore, D. M. (1995). Plasmid-mediated production of class I (AmpC) -lactamase by two Klebsiella pneumoniae isolates from the UK. Journal of Antimicrobial Chemotherapy 35, Chanal, C. M., Poupart, M.-C., Sirot, D. L., Labia, R., Sirot, J. L. & Cluzel, R. A. (1992). Nucleotide sequences of CAZ-2, CAZ-6, CAZ-7 -lactamases genes. Antimicrobial Agents and Chemotherapy 36, Bauernfeind, A., Casellas, J. M., Goldberg, M., Holley, M., Jungwirth, R., Mangold, P. et al. (1992). A new plasmidic cefotaximase from patients infected with Salmonella typhimurium. Infection 20, Gales, A. C., Bolmstrom, A., Jones, R. N., Reis, A. O. & Sader, H. S. (1997). Prevalence, antimicrobial susceptibility and molecular typing of extended-spectrum beta-lactamase-producing Klebsiella pneumoniae from Brazil. In Program and Abstracts of the Thirty- Seventh Interscience Conference on Antimicrobial Agents and Chemotherapy, Toronto, Canada. Abstract D-13, p. 85. American Society for Microbiology, Washington, DC. 30. Jacoby, J. A. & Carreras, I. (1990). Activities of beta-lactam antibiotics against Escherichia coli strains producing extendedspectrum beta-lactamases. Antimicrobial Agents and Chemotherapy 34, Sader, H. S., Mendes, C. F., Pignatari, A. C. & Pfaller, M. A. (1996). Use of macrorestriction analysis to demonstrate interhospital spread of multiresistant Acinetobacter baumannii in Sao Paulo, Brazil. Clinical Infectious Diseases 23, Thomson, K. S., Sanders, W. E. & Sanders, C. C. (1994) USA resistance patterns among UTI pathogens. Journal of Antimicrobial Chemotherapy 33, Suppl. A, Kresken, M., Hafner, D., Mittermayer, H., Verbist, L., Bergogne- Berezin, E., Giamarellou, H. et al. (1994). Prevalence of fluoroquinolone resistance in Europe. Infection 22, Suppl. 2, S Aubert, G., Levy, P. P., Ros, A., Meley, R., Meley, B., Bourge, A. et al. (1992). Changes in the sensitivity of urinary pathogens to quinolones between 1987 and 1990 in France. European Journal of Clinical Microbiology and Infectious Diseases 11, Acar, J. F. & Goldstein, F. W. (1997). Trends in bacterial resistance to fluoroquinolones. Clinical Infectious Diseases 24, Suppl. 1, S Margariti, P. A., Astorri, A. L. & Mastromarino, C. (1997). Urinary tract infections: risk factors and therapeutic trends. Recenti Progressi in Medicina 88, Ouabdesselam, S., Hooper, D. C., Tankovic, J. & Soussy, C. J. (1995). Detection of gyra and gyrb mutations in quinolone-resistant clinical isolates of Escherichia coli by single-strand conformational polymorphism analysis and determination of levels of resistance conferred by two different single gyra mutations. Antimicrobial Agents and Chemotherapy 39, Truong, Q. C., Ouabdesselam, S., Hooper, D. C., Moreau, N. J. & Soussy, C. J. (1995). Sequential mutations of gyra in Escherichia coli associated with quinolone therapy. Journal of Antimicrobial Chemotherapy 36, Received 24 May 1999; returned 13 September 1999; revised 18 October 1999; accepted 26 October

10