High Levels of Antimicrobial Coresistance among Extended-Spectrum-β-Lactamase-Producing Enterobacteriaceae

The production of extended-spectrum β-lactamases (ESBLs) by gram-negative bacteria renders ineffective all penicillins, cephalosporins, and aztreonam in the treatment of serious infections caused by these pathogens (9). Moreover, the phenomenon of antimicrobial coresistance among certain ESBL-producing isolates has been reported, further curtailing the number of drugs useful against these bacteria (1, 11, 12, 16). The literature lacks, however, a comprehensive report of resistance encompassing the Enterobacteriaceae and all potentially useful non-β-lactam antimicrobial agents for both ESBL producers and nonproducers.

At our institution, a 1,200-bed urban teaching hospital, 20% of all bloodstream isolates of Enterobacteriaceae produce ESBLs. This high prevalence must be taken into account when devising empirical treatment regimens for suspected gram-negative infection before culture and susceptibility data are available. Moreover, we know that among the four most common Enterobacteriaceae isolated at our institution, approximately 40% are resistant to trimethoprim-sulfamethoxazole, 30% to ciprofloxacin, 30% to gentamicin, and 15% to piperacillin-tazobactam, while only 1% are resistant to amikacin. We suspected that these high levels of resistance to non-β-lactams are associated with ESBL presence.

We compared antimicrobial coresistance between ESBL-producing and ESBL-nonproducing Enterobacteriaceae to determine the impact of ESBL presence on the likelihood of resistance to antimicrobial classes in addition to β-lactams. We selected cefuroxime-resistant strains of Enterobacteriaceae isolated in the clinical microbiology laboratory from 2000 through 2003 for ESBL testing (8). Two control groups of ESBL nonproducers were assembled from the pool of cefuroxime-resistant (CXM-R) and cefuroxime-nonresistant (CXM-NR) Enterobacteriaceae isolates grown in the microbiology laboratory during the same time period. Urinary isolates were excluded from the study, as NCCLS guidelines for ESBL testing apply to these on an institutional basis only (9). Duplicate isolates from the same patient were excluded.

Confirmatory testing for ESBL presence was performed by disk diffusion using ceftazidime- and cefotaxime-impregnated disks, with and without clavulanic acid (CA) (9). Isolates demonstrating a CA effect, defined as an increase in zone diameter of ≥5 mm in the presence of CA for at least one antibiotic, were considered to be ESBL producers (9).

Antimicrobial susceptibilities were performed using the VITEK 2 system (bioMérieux, Durham, NC) and interpreted according to NCCLS guidelines. Drugs tested were ciprofloxacin, gentamicin, amikacin, piperacillin-tazobactam, trimethoprim-sulfamethoxazole, and imipenem. Organisms intermediately susceptible to a given agent were deemed resistant. The proportion of isolates susceptible to each agent was compared for cases and controls.

Three hundred twelve ESBL producers were surveyed, alongside 929 CXM-NR controls and 287 CXM-R controls. Samples comprised eight different genera. Twelve ESBL producers were resistant to all five antimicrobials tested, while 70 were resistant to four of the agents.

High levels of coresistance (≥40%) were noted among the cases for all agents except amikacin and imipenem. The CXM-NR controls, by contrast, demonstrated ≥89% susceptibility to all agents except trimethoprim-sulfamethoxazole, and still a greater percentage of controls than cases was susceptible to this agent as well (77% versus 30%, P < 0.001 by chi square). Among CXM-R controls, 59% were susceptible to trimethoprim-sulfamethoxazole and ≥65% were susceptible to each of the other agents tested. The results of the case-control comparison are displayed in Table 1. Figure 1 illustrates the burden of resistance to each agent attributable to ESBL presence in each of the four most common Enterobacteriaceae isolated. As is evident, ESBL producers comprise a large proportion of isolates resistant to various antibiotic classes.

In the two decades since ESBLs were first detected, these enzymes have emerged as a major source of antimicrobial resistance among healthcare-associated gram-negative pathogens (16) and are reported in many parts of the world (1, 11, 12, 16). Infection with ESBL-producing pathogens generally necessitates the use of non-β-lactam antibiotics. Data from several studies have suggested that ESBL presence may be associated with resistance to antimicrobial classes other than β-lactams as well, rendering treatment even more difficult (1, 11, 12, 16).

As resistance plasmids are the major source of ESBL transmission (4), and we have determined that the ESBL-producing isolates of common species of Enterobacteriaceae in our institution comprise multiple genotypes (13; unpublished data), we postulate that transferable elements conferring resistance to antimicrobials other than β-lactams travel on or alongside the ESBL-containing plasmids, yielding multidrug-resistant bacteria. Numerous mechanisms exist for resistance to the antimicrobial classes described, and in each case a plasmid-mediated resistance mechanism has been described (6, 7, 10, 15, 17). It is also possible that mechanisms other than, or in addition to, plasmid-mediated cotransfer of various resistance factors account for the phenomenon of coresistance observed (2, 5, 17).

Our findings are worrisome for several reasons. First, our institution has a higher prevalence of ESBLs among isolates of Enterobacteriaceae than that reported in numerous surveys spanning several geographic regions (3, 11, 14, 16). Second, the levels of coresistance among the ESBL isolates are appreciably higher for all antimicrobials tested except amikacin and imipenem, the only two agents with generally adequate coverage of ESBL-producing Enterobacteriaceae. The clinical implications of our findings are that when infection with an ESBL-producing organism is considered a reasonable possibility, appropriate empirical coverage may necessitate some form of combination therapy involving suboptimal agents or may be limited to the carbapenems and amikacin. The dearth of new broad-spectrum antimicrobial agents with activity against gram-negative pathogens further adds to the bleakness of our findings.

The silver lining to our results is the fact that almost all cefuroxime-nonresistant controls, and even the majority of cefuroxime-resistant controls, remained susceptible to all appropriate antimicrobial classes, leaving a broad range of possibilities for effective treatment of infections with these non-ESBL-producing pathogens. If we can bring about a decrease in the prevalence of ESBL producers, the possibilities for empirical coverage of infection with Enterobacteriaceae should become much broader. Thus, if we can control the spread of ESBLs, we can significantly curtail the problem of antimicrobial resistance among nosocomial Enterobacteriaceae isolates. The bleak susceptibility profile offered by the ESBL producers, along with the wide range of treatment options offered by the nonproducers, suggest that appropriate infection control measures aimed at minimizing the spread of ESBLs should be a high priority.

• Open in new tab
• Download powerpoint

FIG. 1.

The burden of antimicrobial resistance attributable to ESBL producers. The bars represent the percentages of ESBL producers among resistant isolates of the four most common Enterobacteriaceae found at our institution, for four agents with activity against gram-negative pathogens. TZP, piperacillin-tazobactam; GM, gentamicin; CIP, ciprofloxacin; SXT, trimethoprim-sulfamethoxazole.

• American Society for Microbiology