- Top of page
This study evaluated the pharmacodynamics of continuous infusion β-lactams against pulmonary isolates of Gram-negative bacteria from patients managed in intensive care units (ICUs) in the USA. Multiple 10 000-patient Monte Carlo simulations were performed by integrating pharmacokinetic data from healthy individuals with 2408 MICs from the 2002 Intensive Care Unit Surveillance System database. These pharmacodynamic simulations suggested that continuous infusion regimens of cefepime, aztreonam, ceftazidime and piperacillin–tazobactam 13.5 g have the greatest likelihood of achieving pharmacodynamic targets against isolates of Enterobacteriaceae in the ICU. β-Lactams are unlikely to achieve pharmacodynamic targets against Pseudomonas aeruginosa or Acinetobacter baumannii when administered as monotherapy.
The threat of antibiotic resistance, coupled with a lack of new antibiotics active against Gram-negative bacteria in the industrial pipeline, necessitates more effective use of existing therapies. β-Lactams are one of the most diverse and clinically useful antibiotic classes, and are active against a wide array of bacterial species in vitro. Unfortunately, in-vitro activity alone is not sufficient to ensure clinical success . Clinical effectiveness depends on the interplay between in-vitro activity, pharmacokinetics, host immune status, tolerability and patient compliance.
In the absence of clinical trial data, pharmacodynamic models satisfy a critical need by enabling the scientific community to predict the likelihood of clinical success based on mathematical models that integrate antimicrobial susceptibility patterns and antimicrobial pharmacokinetics. Animal studies have demonstrated previously that clinical success is best predicted by one of three pharmacodynamic indices: the percentage of time for which the concentration remains above the MIC (%T > MIC); the ratio of peak concentration to MIC (Cmax/MIC); and the ratio of area under the concentration–time curve to MIC (AUC/MIC) . It has been demonstrated previously for β-lactams that animal survival correlates best with %T > MIC [3,4]. This revelation has prompted clinicians to develop dosing strategies that maximise the %T > MIC, including higher doses and shorter dosing intervals [5,6].
While these efforts are noteworthy, intermittent β-lactam administration results in undesirable high peak concentrations and low, potentially sub-MIC, trough concentrations. Administering β-lactams by continuous infusion avoids these fluctuations and enables the %T > MIC to remain above 100% for the entire duration of treatment. When β-lactams are administered by continuous infusion, it can be hypothesised that optimal bactericidal killing is achieved when the steady-state concentration to MIC ratio (Css/MIC) is ≥ 4 [7,8].
The present study evaluated the pharmacodynamics of several β-lactams when administered by continuous infusion against pulmonary isolates of Gram-negative bacteria from patients managed in intensive care units (ICUs) in the USA, with the aim of providing valuable insights into which continuous infusion β-lactam regimens are most likely to treat Gram-negative pulmonary infections in the ICU effectively.
The Intensive Care Unit Surveillance System (ISS) database (Merck & Co., Rathway, NJ, USA) has been described previously . In short, the ISS database is a multi-year, national survey of ICUs in the USA. For the purpose of the present analysis, only MIC data for pulmonary isolates collected during 2002 were evaluated. Bacteria were subdivided into three groups (Enterobacteriaceae, Acinetobacter baumannii and Pseudomonas aeruginosa), and MIC frequency distributions were created for each bacterium/β-lactam combination.
In total, nine continuous infusion β-lactam regimens were modelled, including piperacillin–tazobactam 6.75 g and 13.5 g, piperacillin 6 g and 12 g, ceftazidime 2 g and 3 g, cefepime 3 g and 4 g, and aztreonam 6 g. Pharmacokinetic parameters and their variability were obtained from previous studies of healthy individuals [10–13]. Protein binding data were obtained from the product labelling, and the fraction unbound (fu) was derived by subtracting the % protein binding from 100%.
Crystal Ball 2000 software (Decisioneering, Inc., Denver, CO, USA) was used to create pharmacodynamic models for each bacterium/β-lactam combination, based on published pharmacokinetic data for healthy individuals and MIC distributions from the 2002 ISS database. Free %T > MIC was calculated according to the equation:
where the concentration at steady state (Css, mg/L) is represented by a log-normal distribution, the fraction of unbound drug (fu,%) is represented by a uniform distribution, and the MIC (mg/L) is represented by a discrete distribution.
The 2002 ISS database comprised susceptibility data for 2408 pulmonary isolates of Gram-negative bacteria, including Enterobacteriaceae (n =1430), P. aeruginosa (n = 799) and A. baumannii (n = 179). Table 1 lists the MIC50/90 values for these three bacterial groups of each β-lactam tested.
Table 1. Antibiotic susceptibility of pulmonary isolates of Gram-negative bacteria from the 2002 Intensive Care Unit Surveillance System (ISS) database
|Antibiotic||MIC50/90 (% susceptible)|
| Enterobacteriaceae (n = 1430)||Pseudomonas aeruginosa (n = 799)||Acinetobacter baumannii (n = 179)|
|Piperacillin–tazobactam||4/64 (84)||4/128 (89)||32/128 (48)|
|Piperacillin||4/128 (60)||4/128 (84)||128/128 (34)|
|Ceftazidime||1/32 (85)||2/32 (81)||16/32 (48)|
|Cefepime||0.06/4 (94)||4/32 (75)||16/64 (39)|
|Aztreonam||1/32 (85)||4/32 (67)||32/64 (17)|
Table 2 depicts the results of pharmacodynamic simulations against the three bacterial groups. The probability of achieving a Css/MIC ratio of ≥ 1 for Enterobacteriaceae was ≥ 80% for cefepime, aztreonam, ceftazidime and piperacillin–tazobactam. Neither piperacillin regimen was ≥ 80% likely to achieve a Css/MIC ratio of ≥ 1. When the Css/MIC ratio was raised to ≥ 2, the piperacillin regimens remained at < 80%, and piperacillin–tazobactam 6.75 g was the only additional antibiotic regimen with a probability of < 80%. Furthermore, as the Css/MIC ratio was increased to ≥ 4, piperacillin 6 g, piperacillin 12 g, piperacillin–tazobactam 6.75 g and piperacillin–tazobactam 13.5 g all had a probability of < 80%. Against P. aeruginosa, only ceftazidime, cefepime 4 g and piperacillin–tazobactam 13.5 g achieved probabilities of ≥ 80% at a Css/MIC ratio of ≥ 1. None of the tested β-lactams achieved probabilities of ≥ 80% at a Css/MIC ratio of ≥ 1 against A. baumannii.
Table 2. Pharmacodynamics of continuous infusion β-lactams against pulmonary isolates of Enterobacteriaceae, Pseudomonas aeruginosa and Acinetobacter baumannii from the 2002 Intensive Care Unit Surveillance System (ISS) database
| ||Probabilities (%) that the free Css/MIC ratio will be ≥ 1, ≥ 2 or ≥ 4 |
|Enterobacteriaceae||Pseudomonas aeruginosa||Acinetobacter baumannii|
|≥ 1||≥ 2||≥ 4||≥ 1||≥ 2||≥ 4||≥ 1||≥ 2||≥ 4|
|Aztreonam 6 g||88||85||82||78||63||42||36||14||5|
|Cefepime 3 g||94||91||88||72||51||28||37||28||16|
|Cefepime 4 g||95||93||90||82||65||43||46||35||25|
|Ceftazidime 2 g||86||85||83||80||72||53||46||33||14|
|Ceftazidime 3 g||86||84||83||83||76||63||49||41||24|
|Piperacillin–tazobactam 6.75 g||82||64||7||79||72||54||40||28||3|
|Piperacillin–tazobactam 13.5 g||86||84||76||80||73||57||48||42||33|
|Piperacillin 6 g||58||42||5||63||42||4||20||7||1|
|Piperacillin 12 g||67||60||50||78||67||52||34||22||8|
The rationale behind therapy with continuous infusion β-lactams has been well-described, and previous studies have documented equivalent or improved pharmacodynamics compared with traditional intermittent administration [11,14,15]. The present study is the first comparative evaluation of continuous infusion β-lactam regimens, and offers valuable information regarding the predicted efficacy of β-lactams in the ICU setting. While these studies suggest that the administration of β-lactams by continuous infusion improves the pharmacodynamics, it is less certain whether continuous infusion improves patient outcomes. The available evidence comes largely from five small prospective clinical trials [16–20]. Grant et al. compared the clinical efficacy of continuous vs. intermittent infusion of piperacillin–tazobactam. Days to normalisation of fever were significantly lower in the continuous infusion group (1.2 ± 0.8 days vs. 2.4 ± 1.5 days; p 0.012). Although the difference was not statistically significant, continuous infusion piperacillin–tazobactam outperformed intermittent infusion piperacillin–tazobactam in clinical efficacy (94% vs. 82%; p 0.081) and microbiological success (89% vs. 73%; p 0.092). In addition, two small randomised clinical trials have compared patient outcomes between continuous infusion and intermittent infusion regimens in neutropenic patients, and both demonstrated that outcomes were equal or better with continuous infusion regimens [17,18]. Other studies have demonstrated similar clinical outcomes among immunocompetent patients treated with continuous infusion as opposed to intermittent infusion β-lactams [19,20].
Presently, there is insufficient evidence to warrant the replacement of intermittent infusion by continuous infusion β-lactams as the new standard for all patient populations. Undoubtedly, continuous infusions can be used to maximise β-lactam pharmacodynamics; however, it is still unclear whether continuous infusions will improve patient outcomes, reduce antibiotic resistance, minimise toxicities related to peak concentrations, or decrease healthcare costs.
Overall, the pharmacodynamic simulations in the present study, based on pharmacokinetic data from healthy individuals and national microbiological data, suggest that continuous infusion regimens of cefepime, aztreonam, ceftazidime and piperacillin–tazobactam 13.5 g have the greatest likelihood of achieving pharmacodynamic targets against isolates of Enterobacteriaceae in the ICU. β-Lactams seemed unlikely to achieve pharmacodynamic targets against P. aeruginosa or A. baumannii when administered as monotherapy.