INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

The potent activity of fluoroquinolones (FQs) against a myriad of gram-negative and gram-positive bacterial pathogens has fostered a decade of frequent and continued clinical use. Recently, levofloxacin has broadened the range of indications for FQs to include community-acquired respiratory tract infections attributable to penicillin-resistant Streptococcus pneumoniae. In spite of their success, concern remains regarding the development and increasing prevalence of resistance to FQs among human pathogens and colonizing bacterial species (12, 23, 24, 25).

A decline in the activity of FQs would be especially problematic in view of the ability of gram-negative bacilli to acquire resistance to all other classes of antimicrobials (4, 5, 10, 11, 14, 15, 17, 20, 22, 29, 30). This ability underscores the need to closely monitor FQ activity in the United States and to do so in a timely manner. Recent surveillance studies examining FQ resistance among gram-negative bacilli in the United States have been limited, with published reports focusing largely on bloodstream isolates (7, 18).

To investigate the current status of FQ activity against prominent gram-negative species, as well as any associations with cephalosporin resistance and multidrug resistance, two surveillance strategies were employed. The first was a centralized in vitro study using isolates collected from across the United States and tested at a central reference laboratory. The second strategy utilized The Surveillance Network (TSN) Database-USA, an electronic surveillance network that collects antimicrobial susceptibility data from more than 200 laboratories nationwide. The results from these two strategies were then analyzed and compared.

(This study was presented in part at the 39th Interscience Conference on Antimicrobial Agents and Chemotherapy, San Francisco, Calif., September 1999.)

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Centralized in vitro surveillance study. Between 1 January and 31 March 1999, fresh, prospective clinical isolates of gram-negative bacilli were collected from 27 hospital laboratories geographically distributed throughout the United States. Laboratories were requested to provide defined quotas of a specific species or group of organisms. Isolates were limited to one per patient and were accepted regardless of specimen source, inpatient or outpatient status, or other patient demographic parameters. Upon receipt at the central laboratory, all isolates were subcultured to sheep blood agar, and their identifications were confirmed using the RapID ONE System (Remel, Lenexa, Kans.) for oxidase-negative fermentative gram-negative species and the RapID NF Plus System (Remel) for nonfermentative gram-negative species. In total, 2,684 Enterobacteriaceae isolates, comprised of 204 Citrobacter spp., 183 Enterobacter aerogenes, 323 Enterobacter cloacae, 709 Escherichia coli, 584 Klebsiella pneumoniae, 413 Proteus mirabilis, 72 Providencia spp., and 196 Serratia marcescens isolates, and 684 non-Enterobacteriaceae isolates, including 464 Pseudomonas aeruginosa, 97 Acinetobacter baumannii, and 123 Stenotrophomonas maltophilia isolates, were collected.

Electronic surveillance study. The electronic surveillance study was accomplished using TSN Database-USA. TSN has been operating since 1994 and has been previously described (21; K. M. Tomfohrde, A. V. Mendes, M. L. Hickey, C. Thornsberry, and D. F. Sahm, Progr. Abstr. Int. Conf. Emerg. Infect. Dis., abstr. P-10.7, p. 109, 1998). TSN Database-USA is a repository of quantitative and qualitative antimicrobial susceptibility test results collected from 229 clinical microbiology laboratories throughout the United States. All participant laboratories use commercial or standard susceptibility testing methods. After passing several internal quality control and processing filters, the data is analyzed using a variety of query applications.

Data analysis. Isolates were assessed as susceptible, intermediate, or resistant to each agent tested as defined by NCCLS breakpoint criteria (16). To examine any associated resistance between FQs and extended-spectrum cephalosporins, isolates were also analyzed for ceftazidime susceptibility. The prevalence of multidrug resistance was also investigated. For E. coli and P. mirabilis, a multidrug-resistant (MDR) phenotype was defined as resistance to three or more of the following agents: ampicillin, gentamicin, SXT, and levofloxacin. For all other species of Enterobacteriaceae, resistance to three or more of ceftazidime, gentamicin, SXT, and levofloxacin defined an MDR phenotype. For P. aeruginosa, multidrug resistance included resistance to three or more of ceftazidime, gentamicin, imipenem, and levofloxacin. In all definitions of multidrug resistance, levofloxacin was arbitrarily chosen as the marker drug for FQ resistance. In determining rates of concurrent resistance to FQs and ceftazidime and of multidrug resistance among isolates from TSN, only data from isolates tested against all antimicrobial agents included in the centralized in vitro study were included. This was done to facilitate a balanced comparison of data by the two surveillance methods.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

The activities of all antimicrobials tested against Enterobacteriaceae, P. aeruginosa, A. baumannii, and S. maltophilia, as determined by centralized in vitro testing and TSN, are shown in Table 1. Enterobacteriaceae other than E. coli and P. mirabilis demonstrated ampicillin resistance rates approximately 15 to 30% higher by TSN than by centralized in vitro testing. It is likely that this difference reflects the reporting practices of clinical laboratories, which commonly report isolates of Enterobacteriaceae, excluding E. coli and P. mirabilis, as resistant to ampicillin regardless of susceptibility results. The overriding of susceptibility results by clinical laboratories is done to discourage the clinical use of ampicillin, as current susceptibility testing methods do not readily detect inducible -lactamases, such as AmpC, that are pervasive among certain clinically relevant Enterobacteriaceae. In our centralized in vitro surveillance study we did not alter data in this manner.

According to the centralized in vitro data, the activities of the two FQs studied were comparable for each species of Enterobacteriaceae studied. In most instances, >90% of the isolates were susceptible to both FQs. Exceptions were E. cloacae (88.2% susceptible to ciprofloxacin) and S. marcescens (89.3% susceptible to ciprofloxacin). Regardless of the FQ examined, susceptibility among Providencia spp. did not exceed approximately 65%.

The TSN data revealed similar FQ activities for the Enterobacteriaceae studied (Table 1). The percent susceptible, intermediate, and resistant results provided by the two methods were usually within 2 to 4% of each other. A notable exception was the Providencia spp., for which the percent susceptible results were approximately 10% lower by TSN data than by the centralized in vitro approach.

By centralized in vitro testing, the activities of levofloxacin (71.3% susceptible isolates) and ciprofloxacin (71.1% susceptible isolates) were nearly identical against P. aeruginosa, and these findings were mirrored by those obtained by TSN for levofloxacin (68.8% susceptible isolates) and ciprofloxacin (71.5% susceptible isolates) (Table 1). Similarly, centralized in vitro and TSN results showed A. baumannii at about 50% susceptibility to either of the two agents. However, for levofloxacin the percentage of resistant isolates was notably higher (44.0%) by TSN results than by centralized in vitro testing (32.0%) due to the higher percentage of intermediate isolates by centralized in vitro testing than by TSN (14.4% versus 2.8%). For S. maltophilia, the activity of levofloxacin (88.6% susceptible) was substantially higher than that of ciprofloxacin (34.1% susceptible). The higher activity of levofloxacin than ciprofloxacin was also noted in TSN data, with 78.2 and 28.9% of the isolates being susceptible, respectively (Table 1).

Figure 1 depicts FQ MIC distributions for isolates of P. aeruginosa collected by the centralized in vitro study. The data indicate that at the intermediate breakpoints, levofloxacin (4 µg/ml) and ciprofloxacin (2 µg/ml) inhibited 78.5 and 76.3% of P. aeruginosa isolates, respectively. It was also observed that isolation rates of P. aeruginosa with ciprofloxacin (4.7%) and levofloxacin (5.1%) MICs of >32 µg/ml were similar.

View larger version (29K): [in this window] [in a new window]   FIG. 1.   Distribution of FQ MICs for 464 isolates of P. aeruginosa collected from across the United States during a 1999 centralized in vitro surveillance study.

To examine the association between FQ resistance and resistance to -lactam agents, the activities of levofloxacin and ciprofloxacin were assessed in relation to the ceftazidime susceptibility status of the isolates. For the centralized in vitro study, the percent susceptibility to the FQs was lower in the ceftazidime-nonsusceptible group than in the ceftazidime-susceptible group for every organism group (Table 2). In general, the decrease in percent susceptibility was comparable for both FQs. The most apparent anomaly for this generalization appeared among the eight isolates of ceftazidime-nonsusceptible S. marcescens; however, the number of isolates available in this instance was too small to establish a definitive correlation.

View this table: [in this window] [in a new window]   TABLE 2.   Correlation between FQ and -lactam susceptibilities using centralized in vitro surveillance data

The extent of the decrease in FQ activities among ceftazidime-nonsusceptible groups did vary notably between certain bacterial species. For example, among K. pneumoniae isolates, FQ susceptibility dropped by approximately 40% or more among ceftazidime-nonsusceptible isolates, whereas among E. cloacae isolates, the decrease was only between 10% (for levofloxacin) and 20% (for ciprofloxacin). As noted with the Enterobacteriaceae, a marked reduction in FQ susceptibility also occurred among P. aeruginosa isolates that were not susceptible to either ceftazidime or imipenem. These isolates exhibited a 30 to 40% decrease in susceptibility to both FQs (Table 2).

Using TSN data to examine the association between FQ susceptibility and ceftazidime susceptibility allowed comparisons to be made using considerably larger numbers of isolates for each species (Table 3). However, the same pattern of lower FQ susceptibility among the ceftazidime-nonsusceptible groups was observed. Both levofloxacin and ciprofloxacin activities were comparably decreased among ceftazidime-nonsusceptible groups, and the extent of the decrease varied notably between certain species. For example, as was also evident from the centralized in vitro data in Table 2, FQ susceptibilities were more markedly reduced in the ceftazidime-nonsusceptible isolates of K. pneumoniae than of E. cloacae. Also, as observed with the centralized in vitro study analysis, a 30 to 40% reduction in susceptibility occurred with ceftazidime-nonsusceptible and imipenem-nonsusceptible P. aeruginosa relative to susceptible isolates.

View this table: [in this window] [in a new window]   TABLE 3.   Correlation between FQ and -lactam susceptibilities using electronic (TSN) surveillance data

The percentages of MDR isolates and the most prevalent MDR phenotypes encountered for each species, as reported by TSN and the centralized in vitro study, are presented in Table 4. For all Enterobacteriaceae considered together, TSN identified 2,655 (4.0%) of 67,016 isolates as MDR, while the centralized in vitro testing identified 86 (3.2%) of 2,684 isolates as MDR. With the exception of E. coli, E. cloacae, and S. marcescens, the percentage of MDR isolates was higher in TSN data than in the centralized in vitro study. According to TSN data, multidrug resistance among Enterobacteriaceae occurred most frequently among isolates of K. pneumoniae (7.3%), Providencia spp. (6.0%), and E. cloacae (5.9%). For K. pneumoniae and E. cloacae, the most prevalent MDR phenotype was resistance to ceftazidime, gentamicin, and SXT. For Providencia spp., the most prevalent MDR phenotype was resistance to gentamicin, SXT, and an FQ. Multidrug resistance was less frequently encountered among S. marcescens (0.6%), E. aerogenes (2.5%), and E. coli (3.1%) isolates. Overall, FQ resistance was part of the most prevalent MDR phenotype for all species of Enterobacteriaceae studied except K. pneumoniae and E. cloacae.

The centralized in vitro study showed that multidrug resistance occurred most frequently with E. cloacae (6.5%) and E. coli (4.2%). The most frequent MDR phenotype for E. cloacae was resistance to ceftazidime, gentamicin, and SXT, which matched the most prevalent MDR profile found by TSN. For E. coli, the prominent MDR phenotype was resistance to ampicillin, SXT, and levofloxacin. Multidrug resistance was not encountered among isolates of E. aerogenes and was uncommon among Citrobacter spp. (0.5%) and S. marcescens (1.0%) isolates. In contrast to the TSN data, predominant MDR phenotypes that included an FQ occurred for only two species: E. coli and Providencia spp. Resistance only to levofloxacin was not encountered in any of the Enterobacteriaceae isolates studied by either TSN or the centralized in vitro method.

For P. aeruginosa, multidrug resistance was identified in 6.3% of isolates from TSN and 3.7% of isolates from the centralized in vitro study (Table 4). In TSN data, the most prevalent MDR phenotype was resistance to gentamicin, imipenem, and an FQ (28.3% of MDR isolates) but was followed closely by an MDR phenotype (25.2% of MDR isolates) that included resistance to ceftazidime, gentamicin, imipenem, and an FQ. This MDR phenotype was also the most prevalent among P. aeruginosa isolates tested in the centralized in vitro study. Resistance to an FQ alone was not encountered among isolates of P. aeruginosa by either TSN or centralized in vitro surveillance.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

The ability of clinically relevant gram-negative bacilli to develop resistance to FQs and the potential for this resistance to increase in prevalence underscore the need to monitor resistance trends (2, 5, 8, 9, 10, 19, 27, 29, 30). However, in recent years representative surveillance studies examining FQ resistance among gram-negative bacteria in the United States have been infrequent (7, 13, 18). In the current study, two different surveillance strategies, electronic surveillance using TSN and centralized in vitro testing, were used to examine the current status of FQ activity against commonly encountered species of Enterobacteriaceae and nonfermentative gram-negative bacilli.

The findings of this study are noteworthy from several perspectives. First, for Enterobacteriaceae, both surveillance approaches yielded results that were largely similar, with susceptibility to levofloxacin and ciprofloxacin at approximately 90% or greater for most of the species studied. Second, the results of both approaches were also similar for the nonfermentative gram-negative species studied. The activities of both FQs against P. aeruginosa, A. baumannii, and S. maltophilia were substantially lower than the activities against the Enterobacteriaceae. Third, regardless of the surveillance method used, FQ resistance was notably higher among ceftazidime-resistant Enterobacteriaceae and ceftazidime- and imipenem-resistant P. aeruginosa. Also, multidrug resistance was evident in every species of enteric bacilli and in P. aeruginosa, but the percentage of strains exhibiting multidrug resistance varied among species, as did the most prevalent MDR phenotypes. Finally, resistance to levofloxacin in the absence of resistance to other antibiotics was not encountered among either Enterobacteriaceae or P. aeruginosa.

By both TSN and the centralized in vitro study, Enterobacteriaceae exhibited greater than 90% susceptibility to levofloxacin and ciprofloxacin, and the activities of the two agents were comparable. These findings are consistent with other recent reports that have primarily focused on blood culture isolates collected during 1997 (7, 13, 18). For example, in previous studies E. coli susceptibility to levofloxacin and ciprofloxacin exhibited a narrow range, from 97.2 to 97.6% (7, 13, 18). In the present study, E. coli susceptibilities were slightly lower but still demonstrated a narrow range, from 94.8% (ciprofloxacin) to 95.1% (levofloxacin) by centralized in vitro surveillance and from 96.5% (levofloxacin) to 97.4% (ciprofloxacin) according to TSN (Table 1). The susceptibility rates of other species, such as K. pneumoniae, Enterobacter spp., Citrobacter spp., and S. marcescens, found by TSN and the centralized in vitro study were also consistent with those of earlier studies, with percentages usually ranging from 90 to 96% susceptible (7, 8, 13, 18).

Interestingly, the current overall activities of these FQs against the species of Enterobacteriaceae included in the present study are not substantially different from those reported for ciprofloxacin in a report by Thornsberry and coworkers that included isolates from 1990 and 1991 (23). In that study, ciprofloxacin susceptibility ranged from 91.9 to 99.5%. Therefore, even though FQ resistance can vary between institutions, and certainly among different countries, the findings of this study indicate that in the United States the overall activity of these agents against Enterobacteriaceae has remained consistently high during the 1990s (2, 9, 11, 19, 26, 27). The exception to this finding is the abated activity that levofloxacin and ciprofloxacin demonstrated against Providencia spp. Although the reason for the reduced activity against this particular group of organisms is unknown, it has been reported in other studies and may be related to intrinsic permeability or mutational idiosyncrasies in this genus (3, 6, 19, 23, 28, 29, 30).

With regard to the nonfermentative gram-negative species, the activities of levofloxacin and ciprofloxacin against P. aeruginosa were nearly identical in the TSN and centralized in vitro studies (Table 1). However, the percentage of susceptible isolates (approximately 70% for both FQs) was substantially lower than those reported in previous studies conducted in 1997, in which susceptibilities were 85% for levofloxacin and 89% for ciprofloxacin (7, 18). The lower percent susceptibility found in this study may reflect increasing FQ resistance among P. aeruginosa isolates in the United States. However, other factors, such as the geographic source of the isolates and the fact that the studies reported by Pfaller et al. (18) and Diekema et al. (7) involved only blood isolates, may also have contributed to these differences. In the centralized in vitro study, it was also observed that even though ciprofloxacin tended to be one doubling dilution more potent than levofloxacin at lower concentrations (MICs of 2 µg/ml), the prevalence of P. aeruginosa isolates with ciprofloxacin and levofloxacin MICs of >32 µg/ml was similar, at 4.7 and 5.1%, respectively (Fig. 1). This implies that the cumulative impact of GyrA and ParC mutations, as well as decreases in cellular permeability, conferred similar levels of resistance to both FQs.

The surveillance data presented in this study showed that levofloxacin and ciprofloxacin activities against Acinetobacter spp. were limited, with approximately 50% of the isolates being susceptible to either FQ. These susceptibilities are considerably lower than the 70 to 80% reported in previous studies (7, 13).

Of the gram-negative species included in this study, only S. maltophilia demonstrated marked differences in susceptibility between the FQs. While the percent susceptibility to levofloxacin was well above 70%, ciprofloxacin susceptibility was at least 50% less (Table 1). The poor activity of ciprofloxacin against S. maltophilia has also been documented in previous studies (7, 8, 23).

MDR organisms can present substantial therapeutic challenges, and as such may pose greater public health problems than highly prevalent isolates that exhibit resistance to a single agent. The potential for commonly encountered gram-negative bacilli to acquire cross-resistance to several antimicrobial agents has been well documented (1, 11, 17, 30). For these reasons, and because multidrug resistance has not been commonly addressed in surveillance studies, we investigated the activities of the FQs stratified by ceftazidime susceptibility status (Tables 2 and 3) and also determined the frequencies of MDR phenotypes (Table 4).

Both TSN and centralized in vitro surveillance data demonstrated lower susceptibilities to levofloxacin and ciprofloxacin among ceftazidime-nonsusceptible isolates than among ceftazidime-susceptible ones. Both FQs were comparably affected by concurrent ceftazidime nonsusceptibility. Similar observations were noted for every species examined and were consistent with reports that have evaluated isolates from individual institutions (7, 11, 14, 20, 30). Reductions in FQ activity against ceftazidime-nonsusceptible isolates were most common among K. pneumoniae and E. coli isolates and likely represent clonal spread among these species but may also be due to multifocal emergence. The apparent correlation between FQ resistance and resistance to extended-spectrum cephalosporins is a phenomenon that requires careful monitoring, as such resistant profiles seriously limit the therapeutic options available to treat infections caused by these organisms.

The association between cephalosporin and FQ resistance was also evident when multidrug resistance was investigated (Table 4). Of interest, the percentage of MDR isolates was higher for most species by TSN data than by centralized in vitro study data. Every species exhibited some percentages of isolates that were MDR. The reasons for the differences in multidrug resistance prevalence observed between TSN and centralized in vitro surveillance are uncertain but may include differences in the number of isolates examined, their geographic distribution, or the number of participating institutions. Regardless, multidrug resistance has permeated every one of the most common species of Enterobacteriaceae. Based on reports from individual institutions and from different countries, the prevalence and diversity of MDR phenotypes can substantially expand and become problematic (11, 27); therefore, continued monitoring in the United States is warranted.

Of the MDR phenotypes encountered with the two surveillance approaches, resistance to ceftazidime, gentamicin, and SXT was the most prominent phenotype among isolates of K. pneumoniae and E. cloacae, the two species that had the highest percentage of MDR isolates. However, FQ resistance was part of the most prominent MDR phenotype for all other species studied by TSN. In contrast, FQ resistance was part of the most prominent MDR phenotype only for E. coli and Providencia spp. by the centralized in vitro study. The reasons for the differences between TSN and centralized in vitro surveillance are unclear but again may be a result of differences in the number of isolates examined, their geographic distribution, or the number of institutions participating in the studies.

TSN also reported a higher percentage of MDR isolates for P. aeruginosa than did the centralized in vitro study, and the prominent phenotype differed in that resistance to ceftazidime did not occur. However, in TSN data, ceftazidime resistance was a component of the second most prominent phenotype, which occurred among 25.2% of the MDR isolates.

Resistance only to an FQ was not encountered for any species tested in the study, regardless of surveillance method. While several speculative explanations could be put forth for this observation, it is interesting that resistance to FQs, which are relatively new agents, is most likely to be encountered among organisms resistant to other agents. That is, with the increased use of FQs in a variety of clinical settings, including empiric therapy for outpatients, one might expect resistance to FQs alone to be a phenotype encountered much more frequently than was found in this study.

In summary, both an electronic surveillance network, TSN, and a centralized in vitro study were used to provide comparable and current perspectives on the activities of FQs against several clinically relevant gram-negative bacilli. While this class of agents has maintained an excellent level of activity against clinically relevant Enterobacteriaceae, careful monitoring at local, national, and international levels is still required to provide continuous feedback regarding the resistance status of this important group of organisms. It is imperative to include monitoring the activities of FQs and other antimicrobial classes in the context of associated resistance and multidrug resistance as an important component of any surveillance initiative.