Infections of the blood are extremely serious and require intravenous antibiotic treatment. When the infection results from antibiotic resistant bacteria, the choice of antibiotic is an extremely important decision. Some types of bacteria produce enzymes that may inactivate essential antibiotics, related to penicillin, called 'beta-lactams'. Furthermore high level production of these enzymes can occur during therapy and lead to clinical failure, even when an antibiotic appears effective by laboratory testing. However, this risk of this occurring in clinical practice has only been well described in a limited range of antibiotic classes in a type of bacteria called Enterobacter. There is currently uncertainty as to whether a commonly used, and highly effective antibiotic, called piperacillin-tazobactam is subject to the same risk of resistance developing while on treatment. Infections caused by Enterobacter (and other bacteria with similar resistance mechanisms) are often treated with an alternative drug called meropenem (a carbapenem antibiotic), which is effective but has an extremely broad-spectrum of activity. Excessive use of carbapenems is driving further resistance to this antibiotic class - which represent our 'lastline' of antibiotic defence. As such, we need studies to help us see whether alternatives to meropenem are an effective and safe choice. No study has ever directly tested whether these two antibiotics have the same effectiveness for this type of infection. The purpose of this study is to randomly assign patients with blood infection caused by Enterobacter or related bacteria to either meropenem or piperacillin/tazobactam in order to test whether these antibiotics have similar effectiveness.

Antibiotic resistance is a problem of immense public health significance. Effective antibiotics are essential to complex therapeutic interventions such as transplant medicine, critical care or major surgery. It is estimated that at least 2 million people acquire infections with bacteria that are resistant to standard therapy each year in the United States, with at least 23,000 deaths directly attributable to the infection. With few new antibacterial agents in late-stage development, it has been necessary to consider using existing generic agents in a more targeted approach. This might include revisiting therapeutic options that have previously been considered inferior.

Bloodstream infections caused by Gram negative bacteria are commonly encountered in clinical practice and can be associated with high rates of mortality. Outcomes may be dependent upon the timely administration of appropriate antibiotics, especially in septic shock. Bacteria that possess resistance mechanisms to commonly employed antibiotics are therefore of great concern and may contribute to further mortality.

Over the last 25 years, two antibiotic choices have predominated for intravenous management of these infections. These are combinations of beta-lactam antibiotics with beta-lactamase inhibitors (such as piperacillin/tazobactam) and third generation cephalosporins, such as ceftriaxone. However, some commonly encountered Gram negative organisms possess chromosomally encoded beta-lactamase enzymes, known as AmpC beta-lactamases, that may hydrolyse 3rd generation cephalosporins. Expression of AmpC may be inducible following beta-lactam exposure in some Enterobacteriaceae by loss of inhibitory effects from regulatory elements that control gene transcription. Furthermore, such inducible gene-expression can become constitutively 'de-repressed' by mutational loss of regulatory ampD or ampR genes, leading to high-levels of AmpC production and a phenotype that demonstrates in vitro resistance to most beta-lactams and beta-lactam/beta-lactamase inhibitor (BLBLI) combination agents, except cefepime or carbapenems. Such variants are usually present at low levels (e.g. between 10-5 to 10-7 of the total bacterial population) but may be rapidly selected for during antibiotic therapy.

As a result, AmpC-producing bacteria present particular problems for antibiotic susceptibility reporting and treatment. In vitro susceptibility may not correlate with clinical efficacy as resistance to beta-lactam antibiotics can emerge by selection of variants expressing high levels of AmpC. This has been best described in the context of Enterobacter bacteraemia and therapy with 3rd generation cephalosporins (3GCs). In a landmark study by Chow et al. in 1991, 129 patients with Enterobacter bacteraemia were prospectively examined. Prior cephalosporin use predicted a greater likelihood of identifying a multi-drug resistant isolate on initial blood culture, which was associated with higher subsequent mortality. Furthermore, emergence of resistance to cephalosporins developed during treatment in 6 (19%) of 31 bacteraemic episodes treated with cephalosporins. It is worth noting that this phenomenon was not seen in the small number of patients treated with piperacillin in this study, and that many of the Enterobacter isolates would now be reported as non-susceptible to 3GCs according to current breakpoints. Several other Gram-negative bacteria contain such inducible beta-lactamase genes with the capacity for de-repression. They have been informally labelled the 'ESCPM' group, and are variably described as comprising Enterobacter spp. (especially E. cloacae and E. aerogenes), Serratia marcescens, Citrobacter freundii, Providencia spp. and Morganella morganii.

Clinical studies have shown a variable risk of such emergent resistance and clinical failure occurring with beta-lactam therapy, particularly 3GCs, but when it occurs it has been associated with higher mortality and healthcare-related costs. As a result, 3GCs are usually not recommended as therapy for AmpC-producers, even when susceptible in vitro.

Although few clinical studies have directly addressed this question, carbapenems are often considered optimal therapy for serious infections caused by AmpC producers such as Enterobacter, Serratia or Citrobacter spp. Yet widespread use of carbapenems may cause selection pressure leading to carbapenem-resistant organisms, thus further limiting therapeutic options to "last-line" antibiotics such as colistin or tigecycline. There is therefore a need for establishing the efficacy of generically available alternatives to carbapenems for serious infections caused by bacteria with such AmpC-mediated resistance mechanisms.

Infections caused by ESCPM organisms may also be treated with agents such as quinolones, aminoglycosides, trimethoprim-sulphamethoxazole or cefepime, when susceptibility is proven. However, these have some limitations in terms of toxicity (aminoglycosides), limited contemporaneous efficacy data as well as the adverse effect profile (trimethoprim-sulphamethoxazole) or selective pressure for other multi-resistant organisms or C. difficile (quinolones). A controversial meta-analysis has cast doubt over the safety and efficacy of cefepime, although the significance of this finding has been debated. Beta-lactam/beta-lactamase inhibitor (BLBLI) combination agents, such as piperacillin/tazobactam, have an uncertain role in this context, but are frequently avoided over concerns relating to the development of AmpC-mediated resistance. However, piperacillin-tazobactam, unlike clavulanate-containing BLBLIs, shows some degree of synergy against AmpC de-repressed isolates. In vitro and in animal models, piperacillin-tazobactam appears less able than cephalosporins to select for resistant Enterobacter mutants. Tazobactam is also a less potent inducer of AmpC expression than clavulanate. Furthermore, different 'ESCPM' species display variable degrees of AmpC production; for instance, de-repressed Serratia, Providencia and Morganella strains express levels of AmpC approximately 10-fold below some de-repressed Enterobacter or Citrobacter. It is also worth noting that piperacillin-tazobactam retains activity against M. morganii even when expressing high levels of its AmpC enzyme. It may therefore be misleading to consider 'ESCPM' organisms as a homogenous group in this regard.

The risk of therapeutic failure from the use of BLBLIs for ESCPM organisms that test susceptible has been little studied directly in prospective clinical studies. Retrospective studies would suggest that the risk may be relatively low or even associated with improved outcome. In a study examining 477 patients with Enterobacter bacteraemia, the risk of emergent AmpC-mediated resistance with broad-spectrum cephalosporin therapy was 19% - in concordance with the original finding of Chow et al - and remained a significant risk factor in a multivariate analysis (RR = 2.3; 95% CI 1.2-4.3). However, there was no association with emergent resistance and the use of piperacillin-tazobactam (RR 1.1; 95% CI 0.4-2.7) or other BLBLI combinations, although these agents were not frequently used. A later study analysing 377 consecutive episodes of Enterobacter bacteraemia, the only factor independently associated with a reduction in 30 day mortality was empirical use of piperacillin-tazobactam (OR 0.11; 95% CI 0.01-0.99), although again only 13.1% and 35.4% of patients received this agent as empirical and definitive therapy respectively.

The concept that BLBLIs are to be universally avoided for infections caused by AmpC producers, even when susceptibility is proven, has been questioned. There is great variation in clinical practice and laboratory reporting across Australia and the world in this regard. Demonstrating, in a well-designed clinical trial, that the use of piperacillin-tazobactam for serious infections caused by ESCPM organisms is non-inferior to established options such as carbapenems would prove invaluable to antimicrobial stewardship programs aiming to restrict carbapenem or quinolone use.

We still have relatively few clinical studies to help guide therapeutic decisions for infections caused by AmpC-producers, and no randomised-controlled trials specifically examining this question. Bloodstream infections caused by such bacteria are relatively common and can drive the use of broad-spectrum antibiotic use. Given the alarming emergence of bacterial resistance to 'last-line' antibiotics such as carbapenems, we urgently require well designed studies to guide therapeutic decisions in this area.

Both meropenem and piperacillin-tazobactam are antibiotics that have been widely used in clinical practice for many years. They have proven efficacy in a wide range of infectious syndromes, including severe sepsis, febrile neutropenia, ventilator-associated pneumonia and intra-abdominal sepsis. Both agents are licenced for the treatment of serious infections and are available for routine clinical use in generic form.

Inclusion Criteria:

• Bloodstream infection with Enterobacter spp., Serratia marcescens, Providencia spp., Morganella morganii or Citrobacter freundii (i.e. likely AmpC-producer), and susceptibility to 3rd generation cephalosporins (i.e. ceftriaxone, cefotaxime or ceftazidime), meropenem and piperacillin-tazobactam from at least one blood culture draw. This will be determined in accordance with laboratory methods and susceptibility breakpoints defined by protocols used in the recruiting site laboratories..
• No more than 72 hours has elapsed since the first positive blood culture collection.
• Patient is aged 18 years and over (>=21y in Singapore).

Exclusion Criteria:

1. Patient not expected to survive more than 4 days
2. Patient allergic to a penicillin or a carbapenem
3. Patient with significant polymicrobial bacteraemia (that is, a Gram positive skin contaminant in one set of blood cultures is not regarded as significant polymicrobial bacteraemia).
4. Treatment is not with the intent to cure the infection (that is, palliative care is an exclusion).
5. Pregnancy or breast-feeding.
6. Use of concomitant antimicrobials in the first 4 days after enrolment with known activity against Gram-negative bacilli (except trimethoprim/sulphamethoxazole may be continued as Pneumocystis prophylaxis).
7. Severe acute illness as defined by Pitt bacteraemia score of >4
8. Likely source to be from (proven or suspected at the time of randomisation) the central nervous system, e.g. brain abscess, post-surgical meningitis, shunt infection (due to concerns over CNS penetration of piperacillin/tazobactam)