Point-Counterpoint: Piperacillin-Tazobactam Should Be Used To Treat Infections with Extended-Spectrum-Beta-Lactamase-Positive Organisms

INTRODUCTION

Beta-lactam/beta-lactamase inhibitor combination antimicrobials (BLBLIs) are among the most controversial classes of antibiotic agents available for the treatment of infections caused by extended-spectrum-beta-lactamase (ESBL)-producing Gram-negative bacteria (ESBL-GNR). Piperacillin-tazobactam (PTZ) is one of the most frequently utilized antibiotic agents for empirical Gram-negative bacterial coverage and remains active against a large proportion of ESBL-GNR strains. Furthermore, good antimicrobial stewardship practices encourage the use of carbapenem-sparing treatment regimens for infections due to ESBL-GNR. As rapid diagnostics are increasingly used in the clinical microbiology laboratory and have the capability of detecting CTX-M type or other ESBL resistance mechanisms, this issue continues to be pertinent. Some data imply reduced efficacy of PTZ against ESBLs. Several factors may affect a clinician's choice to use BLBLIs, including the isolate's MIC, the site and severity of infection, and the type of resistance mechanism. These factors are explored in this review of the pros and cons of BLBLI treatment of invasive infections due to ESBL-producing bacteria, as well as how laboratories should report results for BLBLIs for these organisms as they relate to antimicrobial stewardship. In this Point-Counterpoint, Audrey Schuetz provides the pro point of view and Sergio Reyes and Pranita Tamma provide the con, counterpoint view.

POINT

In this era of increasing focus on antimicrobial stewardship, clinicians and laboratorians are asking whether it is worth revisiting use of beta-lactam/beta-lactamase inhibitor combination antimicrobials (BLBLIs) for treatment of infections due to extended-spectrum-beta-lactamase (ESBL)-producing Gram-negative bacteria (ESBL-GNR). Over time, carbapenems have become the first-line treatment option for infections due to ESBL-GNR (1). The increased use of carbapenems has led in part to a rise in the number of infections by carbapenem-resistant Gram-negative bacteria. Literature reporting the clinical efficacy of carbapenems has largely driven this preference. Clinicians and laboratorians alike are currently reassessing the possibility of treatment of ESBLs with piperacillin-tazobactam (PTZ). It is well recognized that the results from the assays used in the past for evaluating in vitro susceptibility to clavulanate as confirmatory tests for ESBL production may not, in fact, translate to in vivo susceptibility and clinical efficacy. However, health care is entering an era of limited antibacterial treatment options and rising antimicrobial resistance. There are fewer and fewer antimicrobial treatment options available, and interest has been renewed in the potential use of antimicrobials which were largely dismissed in the past.

Publications assessing the efficacy of BLBLIs for treatment of urinary tract infections due to ESBL-GNR support the idea of the efficacy of BLBLIs for this application. One of the earliest retrospective observational studies of PTZ treatment for infections due to ESBL-GNR showed clinical success of PTZ therapy in 6/6 patients with urinary tract infections, regardless of the PTZ MIC for the isolate (2). Organisms in that study included ESBL-producing Escherichia coli, Klebsiella pneumoniae, and Klebsiella oxytoca. Support for the use of BLBLI therapy for treatment of infections by ESBL-GNR from urinary sources was also demonstrated in a retrospective study of infections due to ESBL-producing E. coli and K. pneumoniae (3). In that study, 522 infections due to ESBL-GNR were included, the majority (55%) of which were urinary tract infections. The clinical success of noncarbapenem therapy (80% of which was in the form of BLBLI therapy [primarily cefoperazone-sulbactam]) was similar to that of carbapenem therapy, at 79.6% versus 85.7% (P = 0.15). A recent randomized controlled trial comparing PTZ, cefepime, and ertapenem for the treatment of urinary tract infections due to ESBL-producing E. coli also supports the use of PTZ as an effective treatment for such infections when the isolate tests as susceptible (4). A total of 66 patients received either PTZ or ertapenem in that study. The PTZ MICs were in the susceptible range, between 4 and 16 μg/ml (a MIC of ≤16 μg/ml represents the susceptible breakpoint for Enterobacteriaceae according to the Clinical and Laboratory Standards Institute) (5). Clinical success rates were similar between PTZ and ertapenem (31/33 [93.9%] with PTZ and 32/33 [97.0%] with ertapenem; P = 0.50). Microbiological success, defined as failure to recover E. coli on urine culture performed on days 10 to 14 posttreatment, was achieved in 97.0% (32/33) of both treatment groups, and the levels of 28-day mortality were also the same between the treatment groups at 6.1% (2/33).

Thus, the evidence seems fairly strong supporting use of BLBLIs for treatment of ESBL-GNR causing urinary tract infections, provided that the MIC for the organism is within the susceptible range. There is also evidence that some non-urinary source infections may respond to BLBLI therapy when the MIC for the isolate is within the susceptible range. In the study by Gavin et al. mentioned above which had demonstrated successful PTZ therapy for ESBL-GNR urinary tract infections, successful treatment outcome with PTZ was also seen in 10/11 (91%) patients with non-urinary source infections (including blood, sputum, skin and soft tissue, and other sources) when the organisms demonstrated PTZ MICs of ≤16 μg/ml (2). When the PTZ MIC exceeded 16 μg/ml for isolates from non-urinary source infections, the rate of clinical success was 1/5 (20%).

Observational studies comparing carbapenem therapy to BLBLI therapy for bloodstream infections (BSIs) due to ESBL-GNR have shown different results (6, 7). In a retrospective observational study of 11 ESBL-producing Proteus mirabilis bloodstream infections, only 1/4 (25%) BSIs treated with BLBLIs (including PTZ, amoxicillin-clavulanate, or ampicillin-sulbactam) responded to therapy (6). On the other hand, 5/5 BSIs due to non-ESBL-producing isolates responded to BLBLI therapy (P = 0.02). All BSIs treated with a carbapenem responded to therapy, regardless of the presence or absence of ESBL production. An international prospective observational study of K. pneumoniae BSIs in 1996 to 1997 demonstrated that the efficacy of carbapenems was superior to that of noncarbapenem beta-lactam therapy, including BLBLI therapy (7). Of the 49 K. pneumoniae BSI episodes treated with monotherapy which was considered active in vitro, 2/49 (4%) received PTZ therapy, whereas two other patients received ticarcillin-clavulanate therapy. Mortality at 14 days was 3.7% (1/27) with carbapenem therapy but was 50% (2/4) with BLBLI therapy (both patients treated with PTZ died). Although the numbers of patients treated with BLBLIs in these studies were relatively small, BLBLI therapy compared less favorably to carbapenem therapy in the treatment of BSIs due to ESBL-GNR.

Other studies assessing the use of BLBLI therapy for BSI due to ESBL-GNR have shown conflicting results (8–10). Authors of a retrospective observational study of BSIs due to ESBL-GNR compared the mortality of patients treated with BLBLIs to that of patients treated with carbapenems (8). Of the 33 patients treated with BLBLIs (either PTZ or amoxicillin-clavulanate) to which the isolates displayed in vitro susceptibility, 4 (12%) died compared to 1/28 (3.6%) of those who were treated with carbapenems. Despite the trend in mortality differences, use of a BLBLI for therapy was not associated with a significant increase in mortality (odds ratio [OR], 0.55; 95% confidence interval [CI], 0.19 to 1.55). In another retrospective study of ESBL-GNR BSI, empirical treatment with BLBLIs (primarily PTZ in this study) was associated with a higher, but not statistically significantly higher, mortality rate of 38% (6/16 died) compared to no deaths after empirical carbapenem therapy (0/10) (9). In fact, 5/6 PTZ-treated patients who died had isolates with PTZ MICs in the susceptible range of ≤16 μg/ml.

One of the largest and most recent trials comparing PTZ therapy to carbapenem therapy demonstrated poorer outcomes of PTZ therapy in patients with ESBL-GNR bacteremia (10). In that single-center retrospective study of 213 patients, 103 (48%) received PTZ and 110 (52%) received carbapenems. Seventeen (17%) deaths occurred in the PTZ group and 9 (8%) in the carbapenem group. The adjusted risk of death at 14 days for patients who received PTZ therapy was 1.92 times that seen with patients who received carbapenem therapy (95% CI, 1.07 to 3.45). In that study, approximately 44% of BSIs were associated with central lines in both treatment groups, and there was also a high proportion of patients with pneumonia as the source of bacteremia. These differences in sources may be significant in comparison to other BSI studies in which ESBL-GNR strains have been found to arise predominantly from urinary or biliary sources. Such is the case with a large post hoc analysis of six prospective studies of BSI caused by ESBL-producing E. coli comparing BLBLI therapy with carbapenem therapy (11). In the definitive therapy cohort (in which antibiotics were given after susceptibility reports were released), 54 patients received BLBLI (18 PTZ and 36 amoxicillin-clavulanate) and 120 received a carbapenem. Morality rates at 30 days were similar (9.3% for BLBLI and 16.7% for carbapenem; P > 0.2) for the definitive cohort. However, in the empirical treatment group (in which antimicrobial therapy was administered prior to release of susceptibility results), after adjustment for the propensity score in a Cox regression model, BLBLI therapy showed a hazard ratio for increased mortality of 1.14 (CI, 0.29 to 4.40; P = 0.84). Notably, in that study, high-dose PTZ (4.5 g given intravenously [i.v.] every 6 h) was given instead of the standard dosing of 3.375 g i.v. every 6 h. The higher dosage may favor a reduced mortality difference between the groups. Mortality in that study was also associated with nonurinary and nonbiliary sources of bacteremia.

Tied to the debate on treatment of ESBL infections with BLBLIs is the laboratory's role in labeling isolates as ESBLs and the manner in which BLBLIs are reported for ESBLs. First, should the laboratory perform ESBL confirmatory testing? CLSI has stated that it is unnecessary to perform routine ESBL testing if a laboratory is using the current (e.g., lower) breakpoints for cephalosporins and aztreonam. Breakpoints for these antimicrobials were revised in January 2010, and most commercial AST systems have adopted these revised breakpoints. Prior to the change, laboratories confirmed the presence of an ESBL phenotypically, because results obtained for certain cephalosporins, aztreonam, and penicillins had to be edited to “resistant” if the isolate proved to be an ESBL-GNR. Breakpoint-setting organizations have stated that confirmatory testing for ESBLs may still be useful for epidemiological or infection control purposes such as placement of the patient on contact precautions (5). However, with the increasing focus on carbapenem-sparing therapy and the rising concern with respect to BLBLI treatment for some infections caused by ESBLs, it seems advisable to once again perform ESBL confirmatory testing on isolates. A single surrogate marker for ESBL production, such as ceftriaxone resistance, is not sufficient to detect all ESBLs. The ESBLs are a heterogenous group; close to 1,000 ESBLs are estimated to have been identified, and many have different hydrolyzing abilities for different beta-lactams (K. Bush, personal communication) (12). In fact, the heterogeneity of ESBLs is reflected in the manner in which phenotypic confirmatory ESBL testing is performed, with the use of more than one antimicrobial agent to improve the sensitivity of detection. Additionally, results of testing using surrogate antimicrobial agents are imprecise due to the inherent MIC variability in test systems. Variability in MICs over four or more doubling dilutions even in reference broth dilution testing has been noted for ESBLs in particular (13). Given the heterogenous nature of ESBLs and the variability in MICs seen with different antibiotics, as well as the clinical evidence supporting the use of PTZ treatment for some infections due to ESBL-GNR, it would seem prudent to perform ESBL confirmatory testing on suspected isolates.

The second issue in ESBL reporting is the manner in which laboratories report BLBLIs for ESBL-GNR. This issue was debated in the past when ESBLs were first recognized and cases of PTZ treatment failure for ESBL-GNR were reported. At the time, laboratories took a variety of approaches to reporting PTZ, including automatically reporting results as representing resistance; reporting the MIC and interpretive category with a linked comment stating the possibility of inadequacy of PTZ treatment; reporting only the MIC without an interpretation; and, finally, not including PTZ in the report at all. Given the new knowledge that we have gained concerning the adequacy of PTZ treatment of some types of infections due to ESBL-GNR, laboratories should reassess the manner in which they are reporting PTZ and other similar BLBLIs for ESBL-GNR. Some laboratories may consider appending a comment to the PTZ report for ESBL isolates warning of the inadequacy of treatment for certain types of infections such as bloodstream infections. Others may wish to include a PTZ report only for nonsterile sources. Such laboratory decisions should involve infectious diseases practitioners, pharmacists, and the infection control team.

In summary, BLBLI clinical efficacy data for ESBL-GNR therapy are generally drawn from retrospective and observational studies. Such studies show that the choice of therapy depends on the site and severity of the infection. A recent randomized clinical trial supports the use of PTZ for treatment of urinary tract infections due to ESBL-producing E. coli (4). However, randomized controlled trials specifically comparing carbapenems to BLBLIs for treatment of serious infections due to ESBL-GNR are lacking. The data currently available are limited by the relatively low numbers of patients. Although a few studies of bloodstream ESBL infections have demonstrated the absence of significant differences in mortality or clinical outcome, the majority of studies favor carbapenem therapy over BLBLIs for BSI. The single large prospective cohort study by Rodríguez-Baño et al. that failed to demonstrate significant differences in mortality between BLBLI therapy and carbapenem therapy for BSI was based on higher dosing of PTZ (11). On the basis of these clinical data, it is appropriate to consider treatment of urinary tract infections due to ESBL-producing Enterobacteriaceae with PTZ if an isolate tests as susceptible. There is even some evidence, though it is not strong, supporting the use of BLBLI for therapy in treatment of BSIs associated with sources that are urinary or biliary in origin. Finally, laboratories should review both the need for confirmatory testing of ESBLs and the manner in which data from PTZ (or other similar BLBLIs) are reported for such isolates. Clinical and laboratory evidence supports the performance of confirmatory ESBL testing on suspicious organisms in order to guide appropriate clinical therapy. With the increasing pressure to focus on antimicrobial stewardship, it is appropriate that laboratorians and clinicians alike explore options alternative to carbapenems for treatment of infections due to ESBL-GNR.

Audrey N. Schuetz

COUNTERPOINT

With the ongoing global rise in extended-spectrum-beta-lactamase (ESBL) infections and the need to preserve the efficacy of carbapenems (1), the mounting body of clinical evidence indicating that piperacillin-tazobactam (PTZ) is an effective treatment option for patients with invasive infections by ESBL-producing Gram-negative bacteria (ESBL-GNR) is welcome news (2–5). Previously, the inoculum effect—albeit largely confined to experimental data (6–9), coproduction of additional beta-lactamases not effectively inhibited by beta-lactamase inhibitors (10), and concerns regarding inadequate pharmacokinetic-pharmacodynamic drug target attainment with standard beta-lactam/beta-lactamase inhibitor combination (BLBLI) dosing regimens (11, 12)—led to restrained enthusiasm in considering PTZ for the treatment of invasive ESBL-GNR infections. However, when interpreting available data, it is important to note that there are some stark contrasts between observational studies indicating efficacy of PTZ and carbapenems for the treatment of ESBL-GNR bloodstream infections (2–5) and those that suggest that PTZ use results in poorer outcomes (13, 14). These differences should give us pause in considering PTZ for the treatment of invasive ESBL-GNR.

First, the majority of patients in studies supporting the use of PTZ had “low-inoculum” sources of bloodstream infections (e.g., biliary or urinary sources where the bacterial inoculum was anticipated to be ≤10⁵ CFU/ml) (6–9), ranging from approximately 60% to 90% (2–5). This is in contrast to studies showing inferior outcomes with PTZ use where the minority (roughly 15% to 30%) of bloodstream infections were from low-inoculum sources (13, 14). It seems intuitive that, for infections where relieving an obstruction is arguably the most important component of infection management (i.e., biliary sources) or sites where antibiotic concentrations are expected to be particularly high (i.e., urine), antibiotic therapy may not need to be as “aggressive.” These sites of infection are likely more amenable to pathogen eradication than pneumonia, intra-abdominal collections, deep wound infections, or endovascular infections. Second, ≤15% of patients in ESBL-GNR bloodstream studies supporting PTZ use were critically ill (2–5), whereas one-third to over half of patients in studies suggesting suboptimal outcomes with PTZ required intensive care unit (ICU) care (13, 14). Furthermore, studies favoring PTZ generally included isolates with relatively low piperacillin MICs (∼2 μg/ml) (2–5). In contrast, the median piperacillin MICs approached susceptibility breakpoints in studies indicating inferior outcomes with PTZ (13, 14).

The species of ESBL-GNR may also be an important determinant of the effectiveness of BLBLI treatment. Escherichia coli was the predominant pathogen (70% to 100%) in studies demonstrating similar clinical outcomes for PTZ and carbapenems against ESBL-GNR (2–5). This is in contrast to studies in which other Enterobacteriaceae were recovered with equal or greater frequency (13, 14). Tazobactam has been shown to have increased activity against ESBL-producing E. coli compared to ESBL-producing Klebsiella pneumoniae. The addition of tazobactam to ceftolozane yielded MIC₅₀ and MIC₉₀ values of 0.5 and 4 μg/ml for ESBL-producing E. coli and 32 and >32 for ESBL-producing K. pneumoniae isolates (15). In a large, multicenter study, ESBL-producing K. pneumoniae was independently associated with higher mortality than ESBL-producing E. coli (3). Similarly, in an observational study that identified poorer outcomes in the BLBLI group, almost 70% of patients were infected with ESBL-producing K. pneumoniae isolates (14). Disparities in outcomes across studies may be related to inherent differences in the molecular epidemiology associated with these organisms (i.e., bla_CTX-M types are more often associated with E. coli whereas bla_SHV types are more commonly associated with other Enterobacteriaceae). Additionally, it is unknown if there are microbial characteristics or beta-lactamase characteristics (e.g., inhibitor-resistant SHV beta-lactamases, etc.) or other virulence factors present on mobile genetic elements that might contribute to differences in the conduct of ESBL-producing E. coli compared to other ESBL-producing Enterobacteriaceae. Or, perhaps, E. coli may simply be a proxy for urinary sources of bloodstream infections whereas other Enterobacteriaceae may be more representative of complex sites of infection such as intra-abdominal collections.

Synthesizing the available clinical data, although PTZ may be an effective agent for the treatment of invasive ESBL infections in patients who are not critically ill with lower-inoculum infections and lower piperacillin MICs, one cannot infer that PTZ is effective beyond these parameters based on available observational data. Hopefully, lingering questions will be answered by the MERINO study, the first randomized controlled trial to address the issue of the use of meropenem versus PTZ for ESBL bloodstream infections (16).

The issue then arises as to whether ESBL confirmatory testing is value-added in guiding treating decisions. A number of health care facilities have abandoned confirmatory ESBL testing, in accordance with Clinical and Laboratory Standards Institute (CLSI) guidance (17). This has left many clinicians puzzled as to when ESBLs may be produced. Ceftriaxone nonsusceptibility is often used as a proxy for ESBL presence, and when ceftriaxone MICs greater than 1 μg/ml are observed, carbapenem therapy is frequently pursued. While it is true that ESBL producers are likely to have ceftriaxone MICs in the nonsusceptible range, not all Enterobacteriaceae with ceftriaxone MICs in the nonsusceptible range are ESBL producers (18). ESBL confirmatory testing can be helpful by taking the guesswork out of deciding if an isolate is ESBL producing, potentially leading to the avoidance of unnecessary carbapenem therapy. Additionally, the current PTZ CLSI breakpoint is ≤16 μg/ml and the European Committee on Antimicrobial Susceptibility Testing (EUCAST) breakpoint is ≤8 μg/ml. ESBL isolates with PTZ MICs nearing the breakpoints may not respond as favorably to PTZ as isolates with lower PTZ MICs (19, 20). Knowing when Enterobacteriaceae with PTZ MICs in these higher ranges are ESBL producers is helpful to guide clinicians toward alternative regimens. Taking the data together, we believe that performing ESBL confirmatory testing can favorably impact antibiotic-prescribing decisions.

Sergio Reyes and Pranita D. Tamma

SUMMARY

Points of agreement

1. The decision to use PTZ to treat an infection with an ESBL-GNR is complex and requires consideration of the source of the infection, the severity of the infection, the identity of the organism, the MIC of the organism, and the dosage of antibiotic used.

2. PTZ may be effective for treating invasive ESBL-GNR infections in patients who are not critically ill and who have a lower inoculum of infection and a lower MIC (≤2 μg/ml).

3. The strongest data supporting the use of a BLBLI for treating infections caused by ESBL-GNR are those from urinary tract infections and possibly biliary tract infections.

4. Therapy using BLBLIs appears to be less effective than carbapenem therapy for bloodstream infections due to ESBL-GNR.

5. If PTZ is used for these infections, the laboratory should report MIC data and perform ESBL confirmatory testing to provide clinicians with optimal information for clinical decisions.

6. Laboratories that perform ESBL confirmatory testing should consider inclusion of a comment indicating that PTZ therapy may be inadequate for treating bloodstream infections or other serious infections.

Points requiring further consideration

1. The efficacy of BLBLIs for bloodstream infections due to ESBL-producing organisms needs to be assessed in a large multicenter trial. The ongoing MERINO study, a randomized controlled trial of PTZ versus meropenem for the treatment of bloodstream infections due to these organisms, should provide much-needed data on appropriate treatment options.

2. Similar studies are needed for other types of infections, such as intra-abdominal infections.

3. Outcomes studies assessing the clinical utility of rapid molecular methods for detecting ESBL-producing organism are needed to assist clinical laboratories in determining the ideal approach for confirming identification of ESBL-producing organisms.

Angela M. Caliendo, Editor, Journal of Clinical Microbiology

The views expressed in this feature do not necessarily represent the views of the journal or of ASM.