Evaluation and Updating of the Osiris Expert System for Identification of Escherichia coli -Lactam Resistance Phenotypes
http://www.100md.com
微生物临床杂志 2005年第4期
Service de Microbiologie, Hpital Beaujon, Clichy
Bio-Rad, Marnes-la-Coquette
Service de Microbiologie, Hpital Tenon, Paris, France
ABSTRACT
Osiris is a video zone size reader for disk diffusion tests that includes a built-in extended expert system (EES). We evaluated the efficacy of the Osiris EES for the identification of -lactam susceptibility phenotypes of Escherichia coli isolates. Fifteen -lactam agents and three -lactam--lactamase inhibitor combinations were tested by the disk diffusion method against 50 E. coli strains with well-characterized resistance mechanisms. The strains were screened for the production of extended-spectrum -lactamase (ESBL) by the double-disk synergy test using a disk of amoxicillin-clavulanic acid with disks of the extended-spectrum cephalosporins and aztreonam. Overall, the EES accurately identified the phenotype for 78% of the strains, indicated an inexact phenotype for 17% of the strains, and could not find a matching phenotype for the remaining 5% of the strains. The percentage of correct identification for each resistance mechanism was 100% for inhibitor-resistant TEM and for TEM plus cephalosporinase, 88.9% for TEM and for ESBL, 70.8% for cephalosporinase overproduction, and 25% for oxacillinase. The main cause of discrepancy was the misidentification of oxacillinase as inhibitor-resistant TEM. The conventional double-disk synergy test failed to detect ESBL production in two strains (one producing VEB-1 and one producing CTX-M-14), but synergy between cefepime and amoxicillin-clavulanic acid was visible after the distance between the disks was reduced to 20 mm. After the interpretative guidelines of the EES were updated according to our results, the percentage of correct phenotype identification increased from 78 to 96%.
INTRODUCTION
-Lactam resistance in Escherichia coli is predominantly mediated by production of -lactamase. Whereas TEM-1 remains the most frequent enzyme in this species, various types of -lactamases conferring resistance to a wider spectrum of antibiotics have been reported over the last few years (5, 13). First, E. coli isolates resistant to -lactamase inhibitor combinations such as amoxicillin-clavulanic acid were found with a frequency of 5 to 40% in several studies (10, 12, 15, 18). This resistance may be caused by hyperproduction of TEM enzymes, production of inhibitor-resistant TEM (IRT), production of oxacillinases, hyperproduction of the AmpC chromosomal cephalosporinase, or, more rarely, acquisition of plasmid-mediated cephalosporinase (5, 12, 16, 18). Second, extended-spectrum -lactamases (ESBLs) that mediate resistance to extended-spectrum cephalosporins and aztreonam have recently emerged as a major problem in this species (5, 6). In addition to the classical TEM- and SHV-derived enzymes, -lactamases of the CTX-M group and VEB-1 have been increasingly reported (4, 6, 7, 9, 14, 17). Therefore, the routine identification of -lactam resistance phenotypes of E. coli isolates is important for detecting the underlying mechanisms and monitoring the dissemination of the various types of -lactamases.
Osiris (Bio-Rad, Marnes-la-Coquette, France) is a semiautomated system used for the reading and the interpretation of disk diffusion susceptibility tests. It consists of a video camera that measures the inhibition zone diameters and software with a built-in extended expert system (EES). The EES contains a database of interpretative guidelines designed to identify a phenotype matching the susceptibility pattern observed and, if necessary, to change the categorization of certain antibiotics (susceptible, intermediate, or resistant) according to the phenotype detected. For E. coli, it can identify eight -lactam resistance phenotypes corresponding to the most frequently encountered -lactamases (Table 1).
In the present study, we evaluated the efficacy of the Osiris expert system for identifying -lactam phenotypes of E. coli isolates with well-characterized resistance mechanisms.
MATERIALS AND METHODS
Bacterial strains. Fifty E. coli isolates with various -lactam resistance phenotypes were included in the study. The resistance mechanisms were characterized by biochemical and molecular techniques. Acquired plasmid-mediated -lactamases were identified by PCR and sequencing of the PCR product as described elsewhere (17). Hyperproduction of the AmpC cephalosporinase was detected by susceptibility testing on Mueller-Hinton agar in the presence of 300 mg of cloxacillin/liter. The resistance phenotypes of the 50 strains tested were as follows: TEM-1, 9 strains; production of IRT, 6 strains (3 with IRT-2, 2 with IRT-5, and 1 with IRT-6); production of OXA-1, 8 strains; hyperproduction of AmpC, 8 strains; production of TEM-1 and hyperproduction of AmpC, 4 strains; production of ESBL, 15 strains. Seven of the 15 ESBL-producing strains belonged to the TEM group (1 each with TEM-3, TEM-5, TEM-7, TEM-9, TEM-10, TEM-14, and TEM-52), 3 belonged to the SHV group (1 each with SHV-2a, SHV-4, and SHV-5), 4 belonged to the CTX-M group (1 with CTX-M-1, 1 with CTX-M-14, and 2 with CTX-M-15), and 1 produced VEB-1. One of the strains with CTX-M-14 also produced a plasmid-mediated cephalosporinase (CMY-2), and the strain with VEB-1 was an overproducer of AmpC.
Antimicrobial agents. Disks were supplied by Bio-Rad and contained the following antibiotics: ampicillin (10 μg), amoxicillin-clavulanic acid (20 and 10 μg, respectively), ticarcillin (75 μg), ticarcillin-clavulanic acid (75 and 10 μg), piperacillin (75 and 100 μg), piperacillin-tazobactam (75 and 10 μg or 100 and 10 μg), amdinocillin (10 μg), cefalotin (30 μg), cefamandole (30 μg), cefuroxime (30 μg), cefoxitin (30 μg), cefoperazone (30 and 75 μg), cefixime (5 μg), cefotaxime (30 μg), ceftazidime (30 μg), cefepime (30 μg), cefpirome (30 μg), and aztreonam (30 μg).
Susceptibility testing. The strains were tested at three different centers by the disk diffusion method according to the guidelines of the Comite de l'Antibiogramme de la Societe Franaise de Microbiologie (8). Briefly, Mueller-Hinton agar petri dishes (Bio-Rad) were inoculated by flooding with a bacterial cell suspension adjusted to 106 CFU/ml. Disks containing the antibiotics were distributed on the plates according to a predefined scheme by using a disk dispenser. The disks of extended-spectrum cephalosporins (cefotaxime, ceftazidime, cefepime, and cefpirome) and aztreonam were placed next to a clavulanic acid-containing disk (amoxicillin- or ticarcillin-clavulanic acid) for the detection of ESBL according to the double-disk synergy test described by Jarlier et al. (11). After an incubation of 18 h at 37°C, plates were read with the Osiris video system according to the manufacturer's instructions, and the results were interpreted with the EES. Discrepancies between the phenotype detected and the expected phenotype were analyzed in order to identify the precise cause of the misinterpretations, and the EES database was updated accordingly. The new version of the EES was evaluated using 100 strains of E. coli, including the 50 strains previously tested and 50 further strains with well-characterized resistance phenotypes (10 with TEM, 15 with IRT, 6 with OXA, 5 with high-level AmpC, 2 with TEM plus AmpC, and 12 with ESBL).
RESULTS
The percentage of correct phenotype identification was calculated from the results obtained at the three participating centers (150 tests). Overall, the EES identified the exact phenotype in 117 of 150 (78%) cases. Discrepancies between the phenotype detected and the expected phenotype were observed in 26 of 150 (17%) cases. In the remaining seven (5%) cases, the EES could not find a matching phenotype.
The results were highly variable according to the phenotype considered (Table 2). TEM-1 production was correctly interpreted as low-level or high-level penicillinase in eight strains, but it was misidentified as penicillinase plus cephalosporinase in the remaining strain, which was categorized as resistant to amoxicillin-clavulanic acid (Table 3). The penicillinase-plus-cephalosporinase phenotype was correctly identified in all strains. Comparison of these two phenotypes showed that they could be differentiated by susceptibility to cefoxitin. All strains producing only TEM-1 remained susceptible to cefoxitin, whereas those which also hyperproduced AmpC were resistant to this drug.
IRT production was correctly identified in all cases. In contrast, most OXA-1 strains were misidentified as IRT because the strains remained susceptible to cefalotin and/or because the cefepime inhibition zone was slightly above the 6- to 30-mm expected range. Comparison of the susceptibility patterns observed with IRT and OXA showed that none of the antibiotics tested could clearly differentiate between these two phenotypes (data not shown). The best indicator of OXA production was obtained by comparing the cefepime inhibition zone with the ceftazidime inhibition zone. The cefepime zone was smaller than or equal to the ceftazidime zone in most strains with OXA, whereas it was greater than the ceftazidime zone (>2-mm difference) in all strains with IRT.
The EES correctly identified the cephalosporinase phenotype in 71% of cases, but three types of discrepancies were observed (Table 3). First, low-level cephalosporinase was not detected in two strains because they were categorized as susceptible to cefalotin. Second, the EES did not find a matching phenotype for another strain because the cefotaxime and ceftazidime zone sizes were above the 15- to 25-mm expected range. Third, the EES indicated penicillinase plus cephalosporinase at one center for a strain with high-level AmpC because the ticarcillin zone was within the resistant range.
The double-disk synergy test successfully detected all TEM- and SHV-derived ESBLs, CTX-M-1, and CTX-M-15 (Table 4). Cefotaxime, cefepime, and cefpirome were more effective than ceftazidime for the detection of TEM and SHV enzymes (Fig. 1A and B), whereas ceftazidime and aztreonam were the best indicators for CTX-M-15. One strain with CTX-M-14 and CMY-2 was identified as ESBL producing at a single center because of a slightly positive synergy test with cefepime (Fig. 1C), and it was misinterpreted as producing penicillinase plus cephalosporinase at the two other centers. For the strain producing VEB-1 and AmpC, the EES indicated a penicillinase-plus-cephalosporinase phenotype at all three centers because the synergy test failed to detect ESBL production. The two strains that were misinterpreted were tested by the double-disk synergy test with reduced disk distances (20 mm, center to center). VEB-1 was identified in this modified test by a visible synergy between clavulanic acid and cefotaxime, cefepime, or cefpirome, but not ceftazidime, whereas CTX-M-14 was detected only with cefepime and cefpirome.
The interpretation guidelines of the EES were modified according to the results described above (Table 5), and the susceptibility patterns of 100 strains of E. coli, including the 50 strains previously tested and 50 further strains, were interpreted with the updated version. Overall, the rate of correctly identified phenotypes increased from 78 to 96%, the rate of misidentification decreased from 17 to 2.5%, and the absence of phenotype detection decreased from 5 to 1.5%. The percentage of correct identification was 100% for the penicillinase, penicillinase-plus-cephalosporinase, and ESBL phenotypes, 93.9% for the IRT phenotype, 93.3% for the OXA phenotype, and 86.2% for the cephalosporinase phenotype.
DISCUSSION
Overall, the Osiris EES was able to correctly identify the -lactam resistance phenotypes of 78% of the strains tested. However, the results were highly variable according to the mechanism involved, ranging from 100% for IRT or TEM plus cephalosporinase to 25% for oxacillinases. Four main difficulties were observed: (i) the differentiation of IRT and OXA, (ii) the differentiation of high-level TEM and TEM plus cephalosporinase, (iii) the identification of the cephalosporinase phenotype in strains with various levels of AmpC, and (iv) the detection of ESBL in the presence of chromosomal or plasmid-mediated cephalosporinase.
The main cause of misinterpretation was found in OXA-producing strains. IRT and OXA are both characterized by their poor susceptibility to inhibitors of -lactamases, thus conferring resistance to amoxicillin-clavulanic acid, but they differ by the ability of OXA to hydrolyze cefepime more readily than ceftazidime (1, 3). The differentiation of these two types of enzymes in the initial version of the EES was based on susceptibility to cefalotin and on the cefepime inhibition zone, with a cutoff point at 30 mm (Table 1). However, most OXA enzymes were misidentified as IRT because of susceptibility to cefalotin and/or because the cefepime zone was >30 mm. Analysis of susceptibility results showed that the difference between the ceftazidime zone and the cefepime zone may be predictive of OXA production. Taking this indicator into account allowed an increase in the percentage of correct identification from 25 to 93.3%.
Resistance to -lactamase inhibitor combinations may also be related to high-level production of TEM-1, hyperproduction of AmpC, or an association of these two mechanims (12). In our study, the EES indicated a penicillinase-plus-cephalosporinase phenotype for a strain with high-level TEM-1 which was resistant to amoxicillin-clavulanic acid. Cefoxitin was found to be more effective than amoxicillin-clavulanic acid for differentiating these two phenotypes. All strains producing TEM-1 only remained susceptible to cefoxitin, whereas resistance to this drug indicated an association with hyperproduction of AmpC. After inclusion of susceptibility to cefoxitin in the interpretation guidelines, 100% of TEM-1 enzymes were correctly identified by the EES.
The main difficulty in detecting the cephalosporinase phenotype is that the level of AmpC production may be highly variable. Basal AmpC production is associated with a broad range of MICs of both ampicillin and cefalotin and with susceptibility to cefoxitin and extended-spectrum cephalosporins (13). Strains with enhanced AmpC activity are resistant to cefoxitin and may have decreased susceptibility to extended-spectrum cephalosporins. In our study, no matching phenotype could be found by the EES for two strains with slightly increased AmpC production because the cefalotin zone reached the 18-mm susceptibility breakpoint. The phenotype was not identified for another strain that was resistant to cefalotin and cefoxitin but did not have decreased susceptibility to cefotaxime and ceftazidime. Thus, the inhibition zones of these two latter antibiotics may be highly variable depending on the level of AmpC hyperproduction and therefore should not be considered for the detection of this phenotype. After the EES was updated according to this finding, the percentage of correct identification increased from 70.8 to 86.2%.
Detection of ESBL by the Osiris system is based on the double-disk synergy test with a 30-mm distance between the disks (11). False-negative results have been reported with this conventional test when strains harbor additional -lactamases that mask the clavulanic acid effect. Thus, the detection of ESBL in cephalosporinase-producing strains such as Pseudomonas aeruginosa and Enterobacter species may be improved by reducing the distance between the disks to 20 mm (2, 19). ESBL production was successfully detected by the conventional synergy test in 88.9% of our strains. Cefotaxime, cefepime. and cefpirome scored better than ceftazidime and aztreonam for the identification of TEM- and SHV-derived enzymes, but synergy was visible only with ceftazidime and aztreonam for one strain with CTX-M-15. The conventional synergy test failed to detect VEB-1 and CTX-M-14, which were associated with hyperproduction of AmpC and production of CMY-2, respectively. After the distance between disks was reduced to 20 mm, synergy between cefepime and amoxicillin-clavulanic acid was visible with both enzymes. Therefore, this modified synergy test with cefepime should be performed for all multiresistant E. coli strains with a negative conventional-test result. After this guideline was included in the EES, all ESBLs were successfully detected. The possibility of ESBL detection in those multiresistant strains may also be increased by a double-disk synergy test on Mueller-Hinton agar containing an AmpC inhibitor such as cloxacillin.
In conclusion, this study permitted evaluation of the efficacy of the Osiris system for the identification of E. coli -lactam resistance phenotypes and investigation of the causes of misinterpretation. After the interpretative guidelines of the EES were updated according to our results, the performance of the system was significantly improved. The Osiris EES is a powerful tool for the routine identification of -lactam phenotypes of E. coli isolates producing different -lactamases.
REFERENCES
Aubert, D., L. Poirel, J. Chevalier, S. Leotard, J.-M. Pages, and P. Nordmann. 2001. Oxacillinase-mediated resistance to cefepime and susceptibility to ceftazidime in Pseudomonas aeruginosa. Antimicrob. Agents Chemother. 45:1615-1620.
Bert, F., Z. Ould-Hocine, M. Juvin, V. Dubois, V. Loncle-Provot, V. Lefranc, C. Quentin, N. Lambert, and G. Arlet. 2003. Evaluation of the Osiris expert system for identification of -lactam phenotypes in isolates of Pseudomonas aeruginosa. J. Clin. Microbiol. 41:3712-3718.
Bert, F., C. Branger, and N. Lambert. 2004. Comparative activity of -lactam agents against Pseudomonas aeruginosa strains with well-characterized ticarcillin-resistance mechanisms. Chemotherapy 50:31-34.
Bou, G., M. Cartelle, M. Tomas, D. Canle, F. Molina, R. Moure, J. M. Eiros, and A. Guerrero. 2002. Identification and broad dissemination of the CTX-M-14 -lactamase in different Escherichia coli strains in the northwest area of Spain. J. Clin. Microbiol. 40:4030-4036.
Bradford, P. A. 2001. Extended-spectrum -lactamases in the 21st century: characterization, epidemiology, and detection of this important resistance threat. Clin. Microbiol. Rev. 14:933-951.
Cao, V., T. Lambert, D. Q. Nhu, H. K. Loan, N. K. Hoang, G. Arlet, and P. Courvalin. 2002. Distribution of extended-spectrum -lactamases in clinical isolates of Enterobacteriaceae in Vietnam. Antimicrob. Agents Chemother. 46:3739-3743.
Chanawong, A., F. H. M'Zali, J. Heritage, J. H. Xiong, and P. M. Hawkey. 2002. Three cefotaximases, CTX-M-9, CTX-M-13, and CTX-M-14, among Enterobacteriaceae in the People's Republic of China. Antimicrob. Agents Chemother. 46:630-637.
Comite de l'Antibiogramme de la Societe Franaise de Microbiologie. 2004. Communique 2004. [Online.] Societe Franaise de Microbiologie, Paris, France. http://www.sfm.asso.fr/.
Girlich, D., L. Poirel, A. Leelaporn, A. Karim, C. Tribuddharat, M. Fennewald, and P. Nordmann. 2001. Molecular epidemiology of the integron-located VEB-1 extended-spectrum -lactamase in nosocomial enterobacterial isolates in Bangkok, Thailand. J. Clin. Microbiol. 39:175-182.
Henquell, C. D., D. Sirot, C. Chanal, C. De Champs, P. Chatron, B. Lafeuille, P. Texier, J. Sirot, and R. Cluzel. 1994. Frequency of inhibitor-resistant TEM -lactamases in Escherichia coli isolates from urinary tract infections in France. J. Antimicrob. Chemother. 34:707-714.
Jarlier, V., M.-H. Nicolas, G. Fournier, and A. Philippon. 1988. Extended broad-spectrum -lactamases conferring transferable resistance to newer -lactam agents in Enterobacteriaceae: hospital prevalence and susceptibility pattern. Rev. Infect. Dis. 10:867-878.
Leflon-Guibout, V., V. Speldooren, B. Heym, and M.-H. Nicolas-Chanoine. 2000. Epidemiological survey of amoxicillin-clavulanate resistance and corresponding molecular mechanisms in Escherichia coli isolates in France: new genetic features of blaTEM genes. Antimicrob. Agents Chemother. 44:2709-2714.
Livermore, D. M. 1995. -Lactamases in the clinical laboratory and clinical resistance. Clin. Microbiol. Rev. 8:557-584.
Moland, E. S., J. A. Black, A. Hossain, N. D. Hanson, K. S. Thomson, and S. Pottumarthy. 2003. Discovery of CTX-M-like extended spectrum -lactamases in Escherichia coli isolates from five U.S. states. Antimicrob. Agents Chemother. 47:2382-2383.
Natsch, S., C. Conrad, C. Hartmeier, and R. Schmid. 1998. Use of amoxicillin-clavulanate and resistance in Escherichia coli over a 4-year period. Infect. Control Hosp. Epidemiol. 19:653-656.
Philippon, A., G. Arlet, and G. A. Jacoby. 2002. Plasmid-determined AmpC type -lactamases. Antimicrob. Agents Chemother. 46:1-11.
Saladin, M., V. T. Cao, T. Lambert, J. L. Donay, J. L. Herrmann, Z. Ould-Hocine, C. Verdet, F. Delisle, A. Philippon, and G. Arlet. 2002. Diversity of CTX-M -lactamases and their promoter regions from Enterobacteriaceae isolated in three Parisian hospitals. FEMS Microbiol. Lett. 209:161-168.
Stapleton, P., P. J. Wu, A. King, K. Shannon, G. French, and I. Phillips. 1995. Incidence and mechanisms of resistance to the combination of amoxicillin and clavulanic acid in Escherichia coli. Antimicrob. Agents Chemother. 39:2478-2483.
Tzelepi, E., P. Giakkoupi, D. Sofianou, V. Loukova, A. Kemeroglou, and A. Tsakris. 2000. Detection of extended-spectrum -lactamases in clinical isolates of Enterobacter cloacae and Enterobacter aerogenes. J. Clin. Microbiol. 38:542-546.(Frederic Bert, Manette Ju)
Bio-Rad, Marnes-la-Coquette
Service de Microbiologie, Hpital Tenon, Paris, France
ABSTRACT
Osiris is a video zone size reader for disk diffusion tests that includes a built-in extended expert system (EES). We evaluated the efficacy of the Osiris EES for the identification of -lactam susceptibility phenotypes of Escherichia coli isolates. Fifteen -lactam agents and three -lactam--lactamase inhibitor combinations were tested by the disk diffusion method against 50 E. coli strains with well-characterized resistance mechanisms. The strains were screened for the production of extended-spectrum -lactamase (ESBL) by the double-disk synergy test using a disk of amoxicillin-clavulanic acid with disks of the extended-spectrum cephalosporins and aztreonam. Overall, the EES accurately identified the phenotype for 78% of the strains, indicated an inexact phenotype for 17% of the strains, and could not find a matching phenotype for the remaining 5% of the strains. The percentage of correct identification for each resistance mechanism was 100% for inhibitor-resistant TEM and for TEM plus cephalosporinase, 88.9% for TEM and for ESBL, 70.8% for cephalosporinase overproduction, and 25% for oxacillinase. The main cause of discrepancy was the misidentification of oxacillinase as inhibitor-resistant TEM. The conventional double-disk synergy test failed to detect ESBL production in two strains (one producing VEB-1 and one producing CTX-M-14), but synergy between cefepime and amoxicillin-clavulanic acid was visible after the distance between the disks was reduced to 20 mm. After the interpretative guidelines of the EES were updated according to our results, the percentage of correct phenotype identification increased from 78 to 96%.
INTRODUCTION
-Lactam resistance in Escherichia coli is predominantly mediated by production of -lactamase. Whereas TEM-1 remains the most frequent enzyme in this species, various types of -lactamases conferring resistance to a wider spectrum of antibiotics have been reported over the last few years (5, 13). First, E. coli isolates resistant to -lactamase inhibitor combinations such as amoxicillin-clavulanic acid were found with a frequency of 5 to 40% in several studies (10, 12, 15, 18). This resistance may be caused by hyperproduction of TEM enzymes, production of inhibitor-resistant TEM (IRT), production of oxacillinases, hyperproduction of the AmpC chromosomal cephalosporinase, or, more rarely, acquisition of plasmid-mediated cephalosporinase (5, 12, 16, 18). Second, extended-spectrum -lactamases (ESBLs) that mediate resistance to extended-spectrum cephalosporins and aztreonam have recently emerged as a major problem in this species (5, 6). In addition to the classical TEM- and SHV-derived enzymes, -lactamases of the CTX-M group and VEB-1 have been increasingly reported (4, 6, 7, 9, 14, 17). Therefore, the routine identification of -lactam resistance phenotypes of E. coli isolates is important for detecting the underlying mechanisms and monitoring the dissemination of the various types of -lactamases.
Osiris (Bio-Rad, Marnes-la-Coquette, France) is a semiautomated system used for the reading and the interpretation of disk diffusion susceptibility tests. It consists of a video camera that measures the inhibition zone diameters and software with a built-in extended expert system (EES). The EES contains a database of interpretative guidelines designed to identify a phenotype matching the susceptibility pattern observed and, if necessary, to change the categorization of certain antibiotics (susceptible, intermediate, or resistant) according to the phenotype detected. For E. coli, it can identify eight -lactam resistance phenotypes corresponding to the most frequently encountered -lactamases (Table 1).
In the present study, we evaluated the efficacy of the Osiris expert system for identifying -lactam phenotypes of E. coli isolates with well-characterized resistance mechanisms.
MATERIALS AND METHODS
Bacterial strains. Fifty E. coli isolates with various -lactam resistance phenotypes were included in the study. The resistance mechanisms were characterized by biochemical and molecular techniques. Acquired plasmid-mediated -lactamases were identified by PCR and sequencing of the PCR product as described elsewhere (17). Hyperproduction of the AmpC cephalosporinase was detected by susceptibility testing on Mueller-Hinton agar in the presence of 300 mg of cloxacillin/liter. The resistance phenotypes of the 50 strains tested were as follows: TEM-1, 9 strains; production of IRT, 6 strains (3 with IRT-2, 2 with IRT-5, and 1 with IRT-6); production of OXA-1, 8 strains; hyperproduction of AmpC, 8 strains; production of TEM-1 and hyperproduction of AmpC, 4 strains; production of ESBL, 15 strains. Seven of the 15 ESBL-producing strains belonged to the TEM group (1 each with TEM-3, TEM-5, TEM-7, TEM-9, TEM-10, TEM-14, and TEM-52), 3 belonged to the SHV group (1 each with SHV-2a, SHV-4, and SHV-5), 4 belonged to the CTX-M group (1 with CTX-M-1, 1 with CTX-M-14, and 2 with CTX-M-15), and 1 produced VEB-1. One of the strains with CTX-M-14 also produced a plasmid-mediated cephalosporinase (CMY-2), and the strain with VEB-1 was an overproducer of AmpC.
Antimicrobial agents. Disks were supplied by Bio-Rad and contained the following antibiotics: ampicillin (10 μg), amoxicillin-clavulanic acid (20 and 10 μg, respectively), ticarcillin (75 μg), ticarcillin-clavulanic acid (75 and 10 μg), piperacillin (75 and 100 μg), piperacillin-tazobactam (75 and 10 μg or 100 and 10 μg), amdinocillin (10 μg), cefalotin (30 μg), cefamandole (30 μg), cefuroxime (30 μg), cefoxitin (30 μg), cefoperazone (30 and 75 μg), cefixime (5 μg), cefotaxime (30 μg), ceftazidime (30 μg), cefepime (30 μg), cefpirome (30 μg), and aztreonam (30 μg).
Susceptibility testing. The strains were tested at three different centers by the disk diffusion method according to the guidelines of the Comite de l'Antibiogramme de la Societe Franaise de Microbiologie (8). Briefly, Mueller-Hinton agar petri dishes (Bio-Rad) were inoculated by flooding with a bacterial cell suspension adjusted to 106 CFU/ml. Disks containing the antibiotics were distributed on the plates according to a predefined scheme by using a disk dispenser. The disks of extended-spectrum cephalosporins (cefotaxime, ceftazidime, cefepime, and cefpirome) and aztreonam were placed next to a clavulanic acid-containing disk (amoxicillin- or ticarcillin-clavulanic acid) for the detection of ESBL according to the double-disk synergy test described by Jarlier et al. (11). After an incubation of 18 h at 37°C, plates were read with the Osiris video system according to the manufacturer's instructions, and the results were interpreted with the EES. Discrepancies between the phenotype detected and the expected phenotype were analyzed in order to identify the precise cause of the misinterpretations, and the EES database was updated accordingly. The new version of the EES was evaluated using 100 strains of E. coli, including the 50 strains previously tested and 50 further strains with well-characterized resistance phenotypes (10 with TEM, 15 with IRT, 6 with OXA, 5 with high-level AmpC, 2 with TEM plus AmpC, and 12 with ESBL).
RESULTS
The percentage of correct phenotype identification was calculated from the results obtained at the three participating centers (150 tests). Overall, the EES identified the exact phenotype in 117 of 150 (78%) cases. Discrepancies between the phenotype detected and the expected phenotype were observed in 26 of 150 (17%) cases. In the remaining seven (5%) cases, the EES could not find a matching phenotype.
The results were highly variable according to the phenotype considered (Table 2). TEM-1 production was correctly interpreted as low-level or high-level penicillinase in eight strains, but it was misidentified as penicillinase plus cephalosporinase in the remaining strain, which was categorized as resistant to amoxicillin-clavulanic acid (Table 3). The penicillinase-plus-cephalosporinase phenotype was correctly identified in all strains. Comparison of these two phenotypes showed that they could be differentiated by susceptibility to cefoxitin. All strains producing only TEM-1 remained susceptible to cefoxitin, whereas those which also hyperproduced AmpC were resistant to this drug.
IRT production was correctly identified in all cases. In contrast, most OXA-1 strains were misidentified as IRT because the strains remained susceptible to cefalotin and/or because the cefepime inhibition zone was slightly above the 6- to 30-mm expected range. Comparison of the susceptibility patterns observed with IRT and OXA showed that none of the antibiotics tested could clearly differentiate between these two phenotypes (data not shown). The best indicator of OXA production was obtained by comparing the cefepime inhibition zone with the ceftazidime inhibition zone. The cefepime zone was smaller than or equal to the ceftazidime zone in most strains with OXA, whereas it was greater than the ceftazidime zone (>2-mm difference) in all strains with IRT.
The EES correctly identified the cephalosporinase phenotype in 71% of cases, but three types of discrepancies were observed (Table 3). First, low-level cephalosporinase was not detected in two strains because they were categorized as susceptible to cefalotin. Second, the EES did not find a matching phenotype for another strain because the cefotaxime and ceftazidime zone sizes were above the 15- to 25-mm expected range. Third, the EES indicated penicillinase plus cephalosporinase at one center for a strain with high-level AmpC because the ticarcillin zone was within the resistant range.
The double-disk synergy test successfully detected all TEM- and SHV-derived ESBLs, CTX-M-1, and CTX-M-15 (Table 4). Cefotaxime, cefepime, and cefpirome were more effective than ceftazidime for the detection of TEM and SHV enzymes (Fig. 1A and B), whereas ceftazidime and aztreonam were the best indicators for CTX-M-15. One strain with CTX-M-14 and CMY-2 was identified as ESBL producing at a single center because of a slightly positive synergy test with cefepime (Fig. 1C), and it was misinterpreted as producing penicillinase plus cephalosporinase at the two other centers. For the strain producing VEB-1 and AmpC, the EES indicated a penicillinase-plus-cephalosporinase phenotype at all three centers because the synergy test failed to detect ESBL production. The two strains that were misinterpreted were tested by the double-disk synergy test with reduced disk distances (20 mm, center to center). VEB-1 was identified in this modified test by a visible synergy between clavulanic acid and cefotaxime, cefepime, or cefpirome, but not ceftazidime, whereas CTX-M-14 was detected only with cefepime and cefpirome.
The interpretation guidelines of the EES were modified according to the results described above (Table 5), and the susceptibility patterns of 100 strains of E. coli, including the 50 strains previously tested and 50 further strains, were interpreted with the updated version. Overall, the rate of correctly identified phenotypes increased from 78 to 96%, the rate of misidentification decreased from 17 to 2.5%, and the absence of phenotype detection decreased from 5 to 1.5%. The percentage of correct identification was 100% for the penicillinase, penicillinase-plus-cephalosporinase, and ESBL phenotypes, 93.9% for the IRT phenotype, 93.3% for the OXA phenotype, and 86.2% for the cephalosporinase phenotype.
DISCUSSION
Overall, the Osiris EES was able to correctly identify the -lactam resistance phenotypes of 78% of the strains tested. However, the results were highly variable according to the mechanism involved, ranging from 100% for IRT or TEM plus cephalosporinase to 25% for oxacillinases. Four main difficulties were observed: (i) the differentiation of IRT and OXA, (ii) the differentiation of high-level TEM and TEM plus cephalosporinase, (iii) the identification of the cephalosporinase phenotype in strains with various levels of AmpC, and (iv) the detection of ESBL in the presence of chromosomal or plasmid-mediated cephalosporinase.
The main cause of misinterpretation was found in OXA-producing strains. IRT and OXA are both characterized by their poor susceptibility to inhibitors of -lactamases, thus conferring resistance to amoxicillin-clavulanic acid, but they differ by the ability of OXA to hydrolyze cefepime more readily than ceftazidime (1, 3). The differentiation of these two types of enzymes in the initial version of the EES was based on susceptibility to cefalotin and on the cefepime inhibition zone, with a cutoff point at 30 mm (Table 1). However, most OXA enzymes were misidentified as IRT because of susceptibility to cefalotin and/or because the cefepime zone was >30 mm. Analysis of susceptibility results showed that the difference between the ceftazidime zone and the cefepime zone may be predictive of OXA production. Taking this indicator into account allowed an increase in the percentage of correct identification from 25 to 93.3%.
Resistance to -lactamase inhibitor combinations may also be related to high-level production of TEM-1, hyperproduction of AmpC, or an association of these two mechanims (12). In our study, the EES indicated a penicillinase-plus-cephalosporinase phenotype for a strain with high-level TEM-1 which was resistant to amoxicillin-clavulanic acid. Cefoxitin was found to be more effective than amoxicillin-clavulanic acid for differentiating these two phenotypes. All strains producing TEM-1 only remained susceptible to cefoxitin, whereas resistance to this drug indicated an association with hyperproduction of AmpC. After inclusion of susceptibility to cefoxitin in the interpretation guidelines, 100% of TEM-1 enzymes were correctly identified by the EES.
The main difficulty in detecting the cephalosporinase phenotype is that the level of AmpC production may be highly variable. Basal AmpC production is associated with a broad range of MICs of both ampicillin and cefalotin and with susceptibility to cefoxitin and extended-spectrum cephalosporins (13). Strains with enhanced AmpC activity are resistant to cefoxitin and may have decreased susceptibility to extended-spectrum cephalosporins. In our study, no matching phenotype could be found by the EES for two strains with slightly increased AmpC production because the cefalotin zone reached the 18-mm susceptibility breakpoint. The phenotype was not identified for another strain that was resistant to cefalotin and cefoxitin but did not have decreased susceptibility to cefotaxime and ceftazidime. Thus, the inhibition zones of these two latter antibiotics may be highly variable depending on the level of AmpC hyperproduction and therefore should not be considered for the detection of this phenotype. After the EES was updated according to this finding, the percentage of correct identification increased from 70.8 to 86.2%.
Detection of ESBL by the Osiris system is based on the double-disk synergy test with a 30-mm distance between the disks (11). False-negative results have been reported with this conventional test when strains harbor additional -lactamases that mask the clavulanic acid effect. Thus, the detection of ESBL in cephalosporinase-producing strains such as Pseudomonas aeruginosa and Enterobacter species may be improved by reducing the distance between the disks to 20 mm (2, 19). ESBL production was successfully detected by the conventional synergy test in 88.9% of our strains. Cefotaxime, cefepime. and cefpirome scored better than ceftazidime and aztreonam for the identification of TEM- and SHV-derived enzymes, but synergy was visible only with ceftazidime and aztreonam for one strain with CTX-M-15. The conventional synergy test failed to detect VEB-1 and CTX-M-14, which were associated with hyperproduction of AmpC and production of CMY-2, respectively. After the distance between disks was reduced to 20 mm, synergy between cefepime and amoxicillin-clavulanic acid was visible with both enzymes. Therefore, this modified synergy test with cefepime should be performed for all multiresistant E. coli strains with a negative conventional-test result. After this guideline was included in the EES, all ESBLs were successfully detected. The possibility of ESBL detection in those multiresistant strains may also be increased by a double-disk synergy test on Mueller-Hinton agar containing an AmpC inhibitor such as cloxacillin.
In conclusion, this study permitted evaluation of the efficacy of the Osiris system for the identification of E. coli -lactam resistance phenotypes and investigation of the causes of misinterpretation. After the interpretative guidelines of the EES were updated according to our results, the performance of the system was significantly improved. The Osiris EES is a powerful tool for the routine identification of -lactam phenotypes of E. coli isolates producing different -lactamases.
REFERENCES
Aubert, D., L. Poirel, J. Chevalier, S. Leotard, J.-M. Pages, and P. Nordmann. 2001. Oxacillinase-mediated resistance to cefepime and susceptibility to ceftazidime in Pseudomonas aeruginosa. Antimicrob. Agents Chemother. 45:1615-1620.
Bert, F., Z. Ould-Hocine, M. Juvin, V. Dubois, V. Loncle-Provot, V. Lefranc, C. Quentin, N. Lambert, and G. Arlet. 2003. Evaluation of the Osiris expert system for identification of -lactam phenotypes in isolates of Pseudomonas aeruginosa. J. Clin. Microbiol. 41:3712-3718.
Bert, F., C. Branger, and N. Lambert. 2004. Comparative activity of -lactam agents against Pseudomonas aeruginosa strains with well-characterized ticarcillin-resistance mechanisms. Chemotherapy 50:31-34.
Bou, G., M. Cartelle, M. Tomas, D. Canle, F. Molina, R. Moure, J. M. Eiros, and A. Guerrero. 2002. Identification and broad dissemination of the CTX-M-14 -lactamase in different Escherichia coli strains in the northwest area of Spain. J. Clin. Microbiol. 40:4030-4036.
Bradford, P. A. 2001. Extended-spectrum -lactamases in the 21st century: characterization, epidemiology, and detection of this important resistance threat. Clin. Microbiol. Rev. 14:933-951.
Cao, V., T. Lambert, D. Q. Nhu, H. K. Loan, N. K. Hoang, G. Arlet, and P. Courvalin. 2002. Distribution of extended-spectrum -lactamases in clinical isolates of Enterobacteriaceae in Vietnam. Antimicrob. Agents Chemother. 46:3739-3743.
Chanawong, A., F. H. M'Zali, J. Heritage, J. H. Xiong, and P. M. Hawkey. 2002. Three cefotaximases, CTX-M-9, CTX-M-13, and CTX-M-14, among Enterobacteriaceae in the People's Republic of China. Antimicrob. Agents Chemother. 46:630-637.
Comite de l'Antibiogramme de la Societe Franaise de Microbiologie. 2004. Communique 2004. [Online.] Societe Franaise de Microbiologie, Paris, France. http://www.sfm.asso.fr/.
Girlich, D., L. Poirel, A. Leelaporn, A. Karim, C. Tribuddharat, M. Fennewald, and P. Nordmann. 2001. Molecular epidemiology of the integron-located VEB-1 extended-spectrum -lactamase in nosocomial enterobacterial isolates in Bangkok, Thailand. J. Clin. Microbiol. 39:175-182.
Henquell, C. D., D. Sirot, C. Chanal, C. De Champs, P. Chatron, B. Lafeuille, P. Texier, J. Sirot, and R. Cluzel. 1994. Frequency of inhibitor-resistant TEM -lactamases in Escherichia coli isolates from urinary tract infections in France. J. Antimicrob. Chemother. 34:707-714.
Jarlier, V., M.-H. Nicolas, G. Fournier, and A. Philippon. 1988. Extended broad-spectrum -lactamases conferring transferable resistance to newer -lactam agents in Enterobacteriaceae: hospital prevalence and susceptibility pattern. Rev. Infect. Dis. 10:867-878.
Leflon-Guibout, V., V. Speldooren, B. Heym, and M.-H. Nicolas-Chanoine. 2000. Epidemiological survey of amoxicillin-clavulanate resistance and corresponding molecular mechanisms in Escherichia coli isolates in France: new genetic features of blaTEM genes. Antimicrob. Agents Chemother. 44:2709-2714.
Livermore, D. M. 1995. -Lactamases in the clinical laboratory and clinical resistance. Clin. Microbiol. Rev. 8:557-584.
Moland, E. S., J. A. Black, A. Hossain, N. D. Hanson, K. S. Thomson, and S. Pottumarthy. 2003. Discovery of CTX-M-like extended spectrum -lactamases in Escherichia coli isolates from five U.S. states. Antimicrob. Agents Chemother. 47:2382-2383.
Natsch, S., C. Conrad, C. Hartmeier, and R. Schmid. 1998. Use of amoxicillin-clavulanate and resistance in Escherichia coli over a 4-year period. Infect. Control Hosp. Epidemiol. 19:653-656.
Philippon, A., G. Arlet, and G. A. Jacoby. 2002. Plasmid-determined AmpC type -lactamases. Antimicrob. Agents Chemother. 46:1-11.
Saladin, M., V. T. Cao, T. Lambert, J. L. Donay, J. L. Herrmann, Z. Ould-Hocine, C. Verdet, F. Delisle, A. Philippon, and G. Arlet. 2002. Diversity of CTX-M -lactamases and their promoter regions from Enterobacteriaceae isolated in three Parisian hospitals. FEMS Microbiol. Lett. 209:161-168.
Stapleton, P., P. J. Wu, A. King, K. Shannon, G. French, and I. Phillips. 1995. Incidence and mechanisms of resistance to the combination of amoxicillin and clavulanic acid in Escherichia coli. Antimicrob. Agents Chemother. 39:2478-2483.
Tzelepi, E., P. Giakkoupi, D. Sofianou, V. Loukova, A. Kemeroglou, and A. Tsakris. 2000. Detection of extended-spectrum -lactamases in clinical isolates of Enterobacter cloacae and Enterobacter aerogenes. J. Clin. Microbiol. 38:542-546.(Frederic Bert, Manette Ju)