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


Polymyxins are the most common class of antibiotics used to treat CR GNB as the cornerstone therapy. Although resistance rates have been increasing in some countries, particularly among Enterobacteriaceae,[92–96] polymyxins are still considered the most active agents against CR GNB.[97] Polymyxin B and colistin are the two polymyxins available for clinical use.[14,98,99] Fortunately, significant advances have been made in the past decade in characterizing the PK and PK/PD of these drugs.[36,100–110] The first recommendations for dosing and dose adjustments in renal impairment have been made empirically without consistent PK data supporting them.[14,99] Additionally, no formal recommendations were available for patients on renal replacement therapy (RRT). Consequently, many patients have likely received suboptimal therapy, particularly those with renal dysfunction and those on RRT.

Both polymyxins differ by a single amino acid.[98,111] However, the PK characteristics of polymyxin B and colistin differ noticeably primarily due to the different pharmaceutical forms in which they are administered. While polymyxin B is administered as its active form (polymyxin B sulfate), colistin is administered as an inactive pro-drug, CMS (also called colistimethate), which leads to different PK behaviors.

The PK/PD index that best correlates with bactericidal activity of polymyxins in monotherapy is the free area under the curve (fAUC)/MIC. This was initially suggested by results from a hollow fiber in vitro PD model study in two strains,[112] and more recently confirmed in more extensive dynamic in vitro infection model studies[113] and two animal models.[114,115] However, although fAUC/MIC has been demonstrated to be the best predictor of bactericidal activity of polymyxins, the fAUC/MIC target value has not yet been defined, as considerable between strain variability in the PK/PD target value exists. A broad range of fAUC/MIC values were associated with stasis (1.57–17.3), or 1-log (5.04–42.1), 2-log (6.61–95.0) and 3-log (53.3–141) bacterial killing at 24 h, in the in vitro infection model and in both murine thigh and lung infection models either with P. aeruginosa or with A. baumannii.[113–115] A 'target' average concentration at steady-state (Css,avg) of 2.5 mg/l colistin in patients was proposed by Garonzik et al.,[102] which was similar to the median Css,avg in the 105 patients receiving physician-selected maintenance doses in the study. The Css,avg of 2.5 mg/l also corresponds to an AUC/MIC of 60, which generally led to a magnitude of effect between stasis and 1-log kill in the murine infection models described earlier. This assumes that the average free fraction (f) is similar in infected mice and patients.[114,115]

There are also two retrospective studies supporting that increasing the AUC may improve clinical outcomes.[42,116] Since the AUC value represents the total exposure to the drug, clinically the AUC can be increased by increasing the daily dose. Consequently, it can be expected that higher daily doses would be associated with improved clinical outcomes. Both retrospective studies assessing dosage regime and outcomes of patients treated with polymyxins have demonstrated such results.[42,116] The first study with CMS, assessing 258 intensive care unit patients, has shown that the overall mortality of patients treated with 3, 6 and 9 million IU/day was 38.6, 27.8 and 21.7%, respectively (p = 0.0011).[42] Higher CMS doses were independently associated with lower mortality in the multivariate analysis.[42] The other study evaluated 276 patients treated with polymyxin B and found that hospital mortality was significantly lower in patients receiving daily doses equal to or higher than 200 mg (2 million IU) of polymyxin B. The hospital mortality rates were 66.4, 66.2 and 47.9% in patients receiving <150 mg/day, ≥150 and <200 mg/day and ≥200 mg/day, respectively (p = 0.03).[116] High doses (≥200 mg/day) were independently associated with lower mortality in multivariate analysis, both in the subgroup of patients with microbiologically documented infections (n = 212) and in patients with bloodstream infections (n = 53).[116]

Colistin is administered as CMS, an inactive pro-drug that needs to be converted in vivo to the active drug colistin.[35,117] However, only a small fraction of CMS is converted to colistin in vivo and this conversion is quite slow.[102,107] Therefore, without loading doses, therapeutic concentrations of colistin are only reached after 48 h of CMS administration.[101,102,107] Thus, loading doses of CMS are required to reach therapeutic concentrations of colistin in the first 12–24 h.[102,107] Even with a CMS loading dose, the required conversion from CMS to colistin means that it likely takes several hours until effective colistin concentrations can be achieved.

In contrast, higher plasma concentrations in relation to steady-state (i.e., 65% of steady-state) are attained after the first polymyxin B dose.[36] If a loading dose of 20.000–25.000 IU is given on day 1 of therapy, 85–87% of the steady-state concentration are reached after the first administration of polymyxin B.[36] So, loading doses are also recommended for polymyxin B, although not mandatory as for CMS, particularly in severely ill patients or in infections by organisms with MICs ≥1.0 mg/l.

Although colistin clearance is mainly by the non-renal route, CMS is predominantly cleared by the kidneys. CMS concentrations increase as creatinine clearance decreases, which results in higher concentrations of CMS to be converted to colistin. Therefore patients with impaired renal function require dose adjustment of CMS.[102] In contrast, patients with normal, but especially those with increased creatinine clearances, such as those in initial phases of sepsis and septic shock, will likely present low concentrations of colistin in plasma with usually recommended doses. This is caused by low concentrations of CMS, which is eliminated by the kidneys, and the consequently low fraction of CMS converted to colistin.[102] This is very problematic, particularly for patients with creatinine clearances above 60–70 ml/min. Therefore, it was proposed that colistin be best used as part of a highly active combination, particularly for patients with good renal function and infections by isolates with MICs >0.5 mg/l.[102]

In contrast, the clearance of polymyxin B is not related to creatinine clearance; therefore dose adjustments are not required in renal dysfunction.[36,103] Although one may consider decreasing the daily dose in cases of renal dysfunction, it will ultimately result in low plasma concentrations with potential negative consequences for clinical and microbiological outcomes.[36] It should be noted that in a retrospective cohort study the benefit of high doses of polymyxin B was maintained regardless of the presence of renal dysfunction during therapy.[116]

In patients under RRT (both continuous and intermittent), both CMS and colistin are partially removed,[102,109,118,119] requiring adjustment of dosage regimens as has been proposed by Garonzik et al..[102] There are less data on the PK of polymyxin B in patients under RRT. Data from two patients showed that only 5–12% of polymyxin B are removed in continuous venovenous hemodialysis, indicating that only minimal, if any, increase in the dose would be necessary.[106]

Considering currently available data on the PK of polymyxins, it can be concluded that there are some PK advantages of polymyxin B over CMS/colistin. With currently recommended dosages, polymyxin B reaches higher serum concentrations than colistin, and these polymyxin B concentrations are reached much more quickly, even without a loading dose, which is recommended but does not seem to be as essential as for CMS. Finally, different brands of CMS have similar elemental compositions, but they lead to different exposures to the microbiologically active formed colistin; this is another complication for adjusting dosage regimens since it seems to be unpredictable.[120]

A potential advantage for CMS lies in the treatment of urinary tract infections. As there is substantial tubular reabsorption of polymyxin B (and also colistin), very low concentrations of polymyxin B or colistin are found in urine.[36,103] In contrast, CMS is highly eliminated by the kidneys without tubular reabsorption, and a large amount of CMS is converted to colistin in urine leading to high urinary concentrations of the latter.[35] Thus, although polymyxin B may be successfully used for the treatment of lower urinary tract infections,[121] CMS might potentially have a higher capacity of sterilization of the urine owing to the higher colistin concentration reached at this site. Table 1 summarizes the major differences between the two polymyxins.