Using β-lactam/β-lactamase inhibitors for infections due to extended-spectrum β-lactamase-producing Enterobacteriaceae to slow the emergence of carbapenem-resistant Enterobacteriaceae

KEYWORDS:

• Antimicrobial resistance
• ESBL
• beta-lactam/beta-lactamase inhibitors
• CRE
• blood stream infections

Infections due to multidrug-resistant gram-negative bacteria are rapidly increasing in both health care and community settings. In particular, extended-spectrum β-lactamase (ESBL)-producing Enterobacteriaceae (ESBL-PE) are responsible for 18% of health care-associated Enterobacteriaceae infections in US acute care hospitals (short and long term) and acute rehabilitation facilities [1]. ESBL-PE have been associated with a number of adverse outcomes, excess risk of overall and infection-related mortality, increased length of stay, delay in appropriate therapy, discharge to chronic care, and higher health-care costs [2]. Tumbarello et al. reported that inadequate empiric antibiotic therapy for ESBL-PE leads to a threefold increase in mortality compared to adequate treatment [3]. As a result of their in vitro activity and stability to ESBLs, carbapenems are generally recommended for ESBL-PE infections [4,5]. However, selection pressure caused by carbapenems has been blamed for the spread of carbapenem-resistant Enterobacteriaceae (CRE) [6,7]. Carbapenem-sparing antibiotic regimens that are effective against ESBL-PE could therefore have a major impact by slowing the dissemination of CRE.

Recently, there has been renewed interest in using β-lactam/β-lactamase inhibitors (βL/βLIs) in the treatment of ESBL-PE infections. Enthusiasm for these agents was dampened in the past by data from animal studies, concerns about inoculum effect, worry that additional β-lactamases present within the bacterial genome were not effectively inhibited by βL/βLIs, and inadequate pharmacokinetic/pharmacodynamic drug target attainment with the standard dosing regimens [8]. The inoculum effect may also be more apparent with piperacillin/tazobactam compared to other βL/βLIs, likely as a consequence of the lesser stability of tazobactam than clavulanate to ESBLs [9]. Indeed, the topic of using βL/βLIs for ESBL-PE remains controversial, with some recent data showing inferiority of piperacillin/tazobactam compared to carbapenems for treatment of ESBL bloodstream infections (BSIs) [10]. In contrast, however, a large multinational, retrospective cohort study found that if the βL/βLIs were active in vitro, they were as effective as carbapenems for both empiric and targeted therapy [11]. The 30-day mortality rate was 17.6% and 9.8% in the empirically treated and targeted therapy βL/βLIs groups, respectively, compared to 20% and 13.9% in the carbapenem groups. Notably, ESBL-producing Klebsiella pneumoniae infection was independently associated with greater mortality than ESBL-producing Escherichia coli. Gudiol et al., in a similar large multinational, retrospective cohort study on neutropenic patients with BSIs due to ESBL-PE, also found that 30-day mortality was comparable between those who received carbapenems and βL/βLIs [12]. However, propensity score matching reduced the empirical therapy cohort from 174 patients to 35 pairs, potentially underpowering this study. A recent meta-analysis that included 14 studies compared βL/βLIs with carbapenems for ESBL-PE BSI [13]. No significant difference was found in mortality for patients treated empirically with carbapenems compared to those who received βL/βLIs (22.1% vs. 20.5%, respectively; relative risk [RR], 1.05; 95% confidence interval [CI], 0.83–1.37; P = 0.241). Moreover, among patients who received definitive therapy, there was again no significant difference between carbapenems and βL/βLIs (RR, 0.62; 95% CI, 0.25–1.52; P < 0.001). Five of the studies in the meta-analysis specifically included ESBL E. coli and ESBL K. pneumoniae but did not provide separate data for each pathogen. When just these five studies were analyzed, mortality was higher in the patients who received βL/βLIs. The authors speculated that βL/βLIs might be inferior to carbapenems for ESBL K. pneumoniae BSIs, a conclusion others have reached as well [14–16]. Antibiotic classes other than βL/βLIs, such as fluoroquinolones and aminoglycosides, can also be carbapenem-sparing options to treat ESBL-PE infections if in vitro susceptibilities are favorable [17].

An important factor that may influence outcomes in ESBL-PE BSIs is the source of the bacteremia. Retamar et al. reported on 11 patients who had an ESBL-producing E. coli BSI due to a urinary tract infection (UTI) and were treated with piperacillin/tazobactam [18]. All survived regardless of the minimum inhibitory concentration (MIC). However, for other sources, 30-day mortality was lower for isolates with an MIC of ≤2 mg/l than for isolates with a higher MIC (0% vs. 41.1%; P = 0.02). This could be explained by the fact that piperacillin/tazobactam achieves a high concentration in the urine, despite the primarily biliary excretion of piperacillin, and may be sufficient to overcome resistance. In contrast, pneumonia is a situation where penetration of antibiotics into a compartment with a high inoculum of bacteria is more difficult to achieve [19]. A BSI due to a central line is likely to be more similar to pneumonia in terms of high inoculum than to a UTI, highlighting a reason to promote early line removal in these cases.

While reducing carbapenem use will likely reduce the selective pressure for emergence of CRE, this is unlikely to be a panacea. Indeed, a more nuanced approach is needed in light of data that show the importance of other risk factors. A meta-analysis that included 16 studies found a higher risk for CRE infections with longer length of hospital stay, admission to an intensive care unit (ICU), prior hospitalization, longer days of ICU stay, transplantation, steroid use, central venous catheter use, mechanical ventilation, presence of a tracheostomy, parenteral nutrition, previous antibiotic use, and exposure to carbapenems, aminoglycosides, glycopeptides, quinolones, and antipseudomonal penicillins [20]. Another risk factor is asymptomatic colonization of the gastrointestinal tract by CRE, which can be a reservoir for transmission and may precede infection [21]. Okamoto et al. showed that in patients admitted to long-term acute care hospitals, exposure to higher colonization pressure, exposure to a carbapenem, and higher Charlson comorbidity index were independent risk factors for K. pneumoniae carbapenemase (KPC)-producing Enterobacteriaceae [22]. These investigators suggested that lowering colonization pressure through isolation of patients colonized with KPC and reducing carbapenem exposure may prevent KPC cross-transmission. Thus, familiarity with these risk factors may be beneficial for infection control practitioners in designing novel strategies against CRE transmission. Efforts should also focus on antibiotic stewardship, with a goal of reducing carbapenem use, as well as overall antibiotic use.

A randomized clinical trial (RCT) that compares βL/βLIs to carbapenems for ESBL-PE is needed to determine which regimen leads to better outcomes. The “meropenem versus piperacillin-tazobactam for definitive treatment of bloodstream infections due to ceftriaxone non-susceptible Escherichia coli and Klebsiella spp” (MERINO) trial is recruiting patients in Australia, New Zealand, and Singapore and will compare piperacillin/tazobactam to meropenem for BSIs due to ESBL-PE [23]. While we await these results, how should patients with suspected or confirmed ESBL-PE infections be treated? Results from recent studies show that for many cases of ESBL-producing E. coli, βL/βLIs are effective when susceptible in vitro and adequately dosed, i.e. keeping pharmacokinetic principles in mind in order to achieve therapeutic drug levels. The evidence for treating ESBL-producing K. pneumoniae with βL/βLIs is mixed, so continuing to use carbapenems in these cases may be prudent, at least until RCT data are available. The reasons for the differences between infections from ESBL-producing E. coli and ESBL-producing K. pneumoniae are uncertain but may be related to (1) the source of infection, (2) virulence factors (e.g. immunogenicity of the K. pneumoniae capsule), (3) specific β-lactamase characteristics (e.g. inhibitor-resistant SHV β-lactamases), (4) variations in regional molecular epidemiology, (5) effectiveness of source control, and (6) host factors. Finally, as the use of newer βL/βLIs with activity against ESBL-PE (e.g. ceftolozane/tazobactam and ceftazidime/avibactam) becomes more widespread, their impact on the spread of CRE will become increasingly evident.

Declaration of interest

R Watkins has received grant support and serves on an advisory board for Allergan. The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.

Additional information

Funding

The manuscript was not funded.