Combination Therapy for Carbapenem-resistant Gram-negative Bacteria

Alexandre P Zavascki; Jurgen B Bulitta; Cornelia B Landersdorfer


Expert Rev Anti Infect Ther. 2013;11(12):1333-1353. 

In This Article

Why Combination Therapy?

There are no data from randomized clinical trials (RCTs) with an adequate sample size indicating that combination therapy is the standard of care for patients infected by CR GNB. The first RCTs assessing combination therapy against CR GNB have been only recently published, both assessing the combination of colistin with rifampicin against CR A. baumannii.[23,24] Neither of these two studies has shown a significant benefit of the combination compared with colistin monotherapy.[23,24]

Indeed, there is clinical evidence indicating that combination therapy may not be superior to monotherapy in the treatment of GNB, including P. aeruginosa, when there is susceptibility to a β-lactam, and this drug is used as the cornerstone antibiotic in combination with another drug.[25–28] However, these meta-analyses include data over several decades, which have been subject to a dramatic increase in bacterial resistance over the years. Findings from these meta-analyses cannot be directly extrapolated to the treatment of MDR, XDR or PDR where a β-lactam is rarely used as the cornerstone therapy, and where often there is only a single agent with in vitro susceptibility available. While monotherapy may be appropriate for patients with less severe infections by susceptible isolates, patients with severe infections and critically ill patients would likely benefit most from rationally optimized combination therapy.

The vast majority of combination therapies were chosen empirically without being rationally optimized based on systematic in vitro and animal studies with subsequent translation to humans that is supported by translational modeling. Latest in vitro infection models and animal studies clearly showed that rationally optimized combination therapies are highly promising. The main microbiological reasons for using antibiotic combinations against CR isolates are to maximize the rate and extent of bacterial killing, prevent re-growth and minimize bacterial resistance. These mechanistic reasons support intermediary clinical benefits and ultimately the final clinical and microbiological outcomes summarized in Figure 1.

Figure 1.

Conceptual basis of combination therapy against carbapenem-resistant Gram-negative bacteria.

The combination of polymyxins with another antibiotic has been first proposed for the treatment of POS GNB to overcome some shortcomings of polymyxins in monotherapy, including the potential for therapeutic failure due to the amplification or emergence of hetero-resistant subpopulations.[29–32] Hetero-resistance describes the scenario where resistant bacterial subpopulation(s) are present at initiation of therapy and eventually cause therapeutic failure, as they are not killed, for example, by polymyxin monotherapy. Hetero-resistance needs to be distinguished from adaptive resistance (also called 'tolerance'), which refers to a transient change of bacterial resistance in response to antibiotic therapy. Adaptive resistance has been found both for polymyxins and aminoglycosides in P. aeruginosa, for example. To minimize the impact of adaptive resistance, longer dosing intervals (i.e., 24 h) were suggested for aminoglycosides.[33,34] However, it is not clear whether once daily dosing of polymyxins minimizes emergence of resistance and thus more research is needed.

Also, polymyxins may only achieve limited bacterial killing against isolates with high minimal inhibitory concentrations (MICs) considering the unbound polymyxin concentrations that are achievable in patients.[35,36] Finally, recent studies suggested that polymyxin monotherapy may be inferior to other drugs in the treatment of GNB and have corroborated the idea that combination therapy is necessary.[37–40]

Concomitantly, tigecycline was used as an alternative agent against polymyxin-resistant CR A. baumannii or Enterobacteriaceae isolates. As this drug has not been recommended in monotherapy for severe infections (see below), a second agent was also commonly added. Finally, many CR Enterobacteriaceae presented MICs for carbapenems within the previous susceptibility range, that is, ≤4 mg/l[41] and susceptibility to aminoglycosides. Therefore, carbapenems were prescribed as the cornerstone antibiotic against these organisms in combination with an aminoglycoside. Thus, combination regimens were first prescribed before unequivocal clinical evidence of superiority of this approach, over monotherapy, was available.

In fact, there is still no clinical evidence clearly demonstrating that combination therapy against CR GNB is superior to monotherapy; not even for infections in particularly difficult pathogens such as CR P. aeruginosa and A. baumannii. Indeed, apart from the recent RCTs with colistin and rifampicin against CR A. baumannii, no other study has primarily assessed the effect of combination therapy on clinical outcomes, neither against P. aeruginosa nor A. baumannii, with the exception of the study by Falagas et al.,[41] that evaluated the role of meropenem in combination with colistin methanesulphonate sodium (CMS)/colistin against MDR GNB (predominantly P. aeruginosa and A. baumannii) infections. Nonetheless, no benefit was demonstrated by adding meropenem to the scheme.[41] A further evaluation of these patients[41] together with additional ones has also not found a statistically significant difference between combination therapy and monotherapy with colistin.[42]

It was only for the treatment of CR K. pneumonia bacteremia that some evidence from observational studies has pointed toward a clearer advantage of combination schemes over monotherapy.[43–45] The lower mortality rates observed in patients receiving combination therapy compared with those treated with monotherapy have encouraged physicians to adopt this practice as the standard of care in the treatment of CR Enterobacteriaceae. Additionally, several authors have compiled data from case series and cohort studies and also concluded that combinations were superior to single drug schemes against CR Enterobacteriaceae, particularly those containing a carbapenem.[8,46–48] The promising results with combination therapy against CR Enterobacteriaceae have been extrapolated to the treatment of CR P. aeruginosa and A. baumannii, although, as stated earlier, no clinical benefit against these organisms has been clearly demonstrated so far.

Pre-clinical Studies

Ultimately, a number of pre-clinical studies strongly support the use of rationally optimized combination dosage regimens against CR GNB. In vitro and animal infection models suggest that antibiotic combination regimens are superior to monotherapy for maximizing bacterial killing and minimizing the emergence of resistance (Figure 1). For the interpretation of in vitro studies, the presence of synergy is not important unless the combination also leads to adequate bacterial killing, minimizes the emergence of resistance or ideally achieves both of these goals.

P. aeruginosa. Most studies in contemporary P. aeruginosa isolates with different resistance phenotypes have been focusing on polymyxin-based combinations. Extensive and synergistic bacterial killing of P. aeruginosa by colistin combined with carbapenems (doripenem and imipenem) was most commonly found in static and dynamic in vitro models[49,50] and in murine infection models.[51] This synergy occurred at clinically relevant polymyxin and carbapenem concentrations. Latest dynamic infection models provided strong evidence for colistin plus doripenem preventing emergence of resistance and achieving substantial killing against a very high inoculum of a colistin-resistant and other isolates.[52] The triple drug combination of polymyxin B, doripenem and rifampicin achieved bactericidal activity against five of five CR P. aeruginosa isolates in static time-kill studies at a normal (i.e., low) inoculum.[53] Polymyxin B combined with supra-physiological concentrations of meropenem or amikacin and the associated triple combination achieve strain-dependent synergy against XDR P. aeruginosa.[54] Latest dynamic in vitro infection models showed that combination therapies of meropenem with tobramycin or levofloxacin achieved rapid and substantial killing and minimized resistance against P. aeruginosa with an overexpressed MexAB-OprM efflux pump.[55,56] This pump is clinically highly important as it effluxes almost all β-lactam antibiotics, including meropenem and doripenem but not imipenem.[57] Similar results were obtained for imipenem plus levofloxacin against a P. aeruginosa ioslate with overexpressed efflux or loss of the OprD outer membrane porin, which confers decreased susceptibility to carbapenems.[58,59] These results are in agreement with the synergistic and considerable killing by β-lactam plus aminoglycoside combinations in static time-kill studies, which was observed at least for a considerable fraction of the tested CR P. aeruginosa isolates in older studies.[60–67]

A. baumannii.In vitro and animal infection models against CR A. baumannii (including MDR, XDR and PDR isolates) suggested promising synergy with substantial killing for a polymyxin combined with rifampicin or a carbapenem.[53,68–72] The colistin plus rifampicin combination provided substantial killing and minimized emergence of resistance in the dynamic hollow fiber in vitro infection model over 10 days[73–75] and in the murine thigh infection model.[51] Other studies suggested synergistic and extensive killing for carbapenem plus sulbactam combinations[70,76–78] and for rifampicin combined with imipenem or sulbactam.[72,79] Only a few studies are available on other combinations against CR A. baumannii such as colistin plus tigecycline[80,81] or minocycline.[82] The latter two studies showed promising activity, but further studies are needed on these combinations. Overall, polymyxin plus rifampicin or a carbapenem as well as two or three drug combinations containing a carbapenem, rifampicin or sulbactam are promising and should be further evaluated in vitro and in vivo against CR A. baumannii. More data on the mechanisms of synergy would be highly valuable to more thoroughly elucidate the mechanistic basis for these combinations.

Enterobacteriaceae. Most in vitro and animal data on antibiotic combinations against CR Enterobacteriaceae applies to CR K. pneumoniae and fewer systematic studies are available for Escherichiacoli. Combinations of a polymyxin plus a carbapenem have shown the most consistently beneficial activity against K. pneumoniae in vitro[83–87] and in murine infection models.[51] The combination of colistin plus imipenem yielded synergistic killing against colistin-susceptible, metallo-β-lactamase-producing K. pneumoniae isolates; however, this combination was less promising against colistin-resistant K. pneumoniae.[87] A study in CR K. pneumoniae and CR E. coli showed >3.5 log10 killing at 24 h for polymyxin B plus doripenem against 4 of 5 tested E. coli strains. However, to achieve at least 2.7 log10 killing at 24 h in 5 of 5 K. pneumoniae strains, the triple drug combination of polymyxin B, doripenem and rifampicin was required.[53] Fosfomycin combined with meropenem achieved synergistic killing in 65% of 17 tested KPC-2 producing K. pneumoniae strains, but more studies are needed on this combination.[88]In vitro checkerboard data suggest synergistic killing for colistin plus rifampicin against KPC-producing K. pneumoniae,[89] and further in vitro time-course studies are warranted. Overall, polymyxin plus carbapenem combinations seem most promising based on the available preclinical data. However, much more antibiotic combination studies using static and dynamic in vitro and animal infection models with CR Enterobacteriaceae are clearly needed to rationally optimize the associated combination therapies.

Optimizing Antimicrobial Prescription in Combination Therapy

In combination regimens, it is of paramount importance that the dosage regimens for both the cornerstone and the adjuvant drug are optimized to achieve relevant PK/PD targets and to maximize efficacy, decrease the potential for resistance emergence and decrease toxicity (Figure 1). Significant advances have been made on the knowledge of PK and PK/PD of 'old' drugs used for the treatment of CR GNB, as increasing clinical experience has been published. The knowledge gained in these studies should be utilized as the basis for rational selection of the antibiotics and dosing regimens to successfully treat CR GNB infections and, therefore, is reviewed later. However, most of the PK/PD approaches developed to date only apply to antibiotic monotherapy and rational approaches to optimize combination regimens are scarce.[55,56,90,91]