Pharmacokinetic / pharmacodynamic target attainment of intravenous β-lactam regimens against Gram-negative bacteria isolated in a Brazilian teaching hospital

Introduction: Monte Carlo simulations have been used for selecting optimal antibiotic regimens for treatment of bacterial infections. The aim of this study was to assess the pharmacokinetic and pharmacodynamic target attainment of intravenous β-lactam regimens commonly used to treat bloodstream infections (BSIs) caused by Gram-negative rod-shaped organisms in a Brazilian teaching hospital. Methods: In total, 5,000 patients were included in the Monte Carlo simulations of distinct antimicrobial regimens to estimate the likelihood of achieving free drug concentrations above the minimum inhibitory concentration (MIC; fT > MIC) for the requisite periods to clear distinct target organisms. Microbiological data were obtained from blood culture isolates harvested in our hospital from 2008 to 2010. Results: In total, 614 bacterial isolates, including Escherichia coli, Enterobacter spp., Klebsiella pneumoniae, Acinetobacter baumannii, and Pseudomonas aeruginosa, were analyzed. Piperacillin/tazobactam failed to achieve a cumulative fraction of response (CFR) > 90% for any of the isolates. While standard dosing (short infusion) of β-lactams achieved target attainment for BSIs caused by E. coli and Enterobacter spp., pharmacodynamic target attainment against K. pneumoniae isolates was only achieved with ceftazidime and meropenem (prolonged infusion). Lastly, only prolonged infusion of high-dose meropenem approached an ideal CFR against P. aeruginosa; however, no antimicrobial regimen achieved an ideal CFR against A. baumannii. Conclusions: These data reinforce the use of prolonged infusions of high-dose β-lactam antimicrobials as a reasonable strategy for the treatment of BSIs caused by multidrug resistant Gram-negative bacteria in Brazil.

Owing to the continued emergence of antimicrobialresistant bacterial strains, achieving therapeutic success with the antimicrobials that are currently available has become a major challenge.This is particularly true for infections caused by Gram-negative bacterial species, for which few agents are effective or are in an advanced stage of development (1) (2) .
Gram-negative bacteria are highly effective in acquiring and expressing genes that encode resistance to antimicrobials, making it diffi cult to treat the infections that they cause (3) .Moreover, these resistant organisms are often responsible for nosocomial infections, including pneumonia and urinary tract and bloodstream infections (BSIs), leading to increased morbidity and mortality, along with longer hospitalizations (1) (2) (3) .Indeed, data from the United States (US) National Healthcare Safety Network show that Gram-negative bacteria are responsible for more than 30% of hospital-acquired infections, and are the predominant causes of ventilator-associated pneumonia (47%) and urinary tract infections (45%) in the US.Similar data were reported in several other countries (4) .
Notably, the prevalence of infections due to Gram-negative bacilli is higher in Latin American than in North American medical centers.Additionally, in recent years, decreased antimicrobial susceptibility among Gram-negative species has been observed in Latin American countries, particularly among Pseudomonas aeruginosa and Acinetobacter baumannii strains isolated in Brazil (5) .
Given the scarcity of new antibiotics, pharmacokinetic and pharmacodynamic (PK/PD) concepts have been utilized to optimize the in vivo exposure of these problematic Gram-negative bacteria to antimicrobial compounds.For example, the administration of different infusions of β-lactam antibiotics comprises a viable alternative for eliminating these organisms.Indeed, infusion is the optimal way to maintain antibiotic serum levels above the minimum inhibitor concentration (MIC) for the target organism, and thereby

METHODS
enhance the effi cacy of treatments (6) (7) .The drawbacks to this type of system, however, include the reduced stability of the drug and the need for an intravenous line to maintain the treatment (6) .Despite these drawbacks, extended infusion has been shown to comprise a valid alternative for obtaining meaningful results (8) (9) .In this context, Monte Carlo simulation (MCS) has been used as a tool for aiding in the selection of optimal antibiotic therapies.Through MCS, it is possible to determine dosing regimens that best match the desired therapeutic targets against bacteria of interest (10) (11) .
The aim of this study was to assess the PK/PD target attainment of intravenous β-lactam regimens commonly used to treat BSIs caused by Gram-negative rod-shaped organisms.

Pharmacokinetic model
Steady-state exposures were determined for each antibiotic regimen using serum pharmacokinetic parameters obtained from recently published population pharmacokinetic studies of infected and/or critically ill adult patients (1) (12) .Briefl y, the PK parameters included body clearance (CL), volume of distribution (Vd), and fraction of unbound (free) drug (f).The methodology used to simulate steady-state antibacterial exposures in a population of adult patients with normal renal function (i.e., ≥50mL/min) has been previously described (13) .

Monte Carlo simulation
A 5,000-patient MCS (Crystal Ball 2000; Decisioneering, Inc., Denver, CO, USA) was conducted to estimate the % fT > MIC ratio for each antibiotic regimen/bacterial population combination, as well as the probability of a simulated patient achieving the pharmacodynamic target [referred to as the probability of target attainment (PTA)].Probability of target attainment was calculated over a range of doubling MICs between 0.5 and 128mg/L.During each interaction, CL (in liters), Vd, f, and MIC values were substituted into the appropriate equations based on the probability distributions, thereby resulting in 5,000 different estimates of pharmacodynamic exposure for each antibiotic regimen tested against each bacterial species.
PTAs were then used to calculate the cumulative fraction of response (CFR) for each antibiotic regimen against each bacterial population.CFRs were calculated as the summation of PTAi*Fi, with the subscript i indicating the MIC category, ranked from the lowest to the highest MIC value for a population of microorganisms, PTAi denoting the PTA of each MIC category for that drug regimen, and F denoting the fraction of the population of microorganisms at each MIC category.Regimens that achieved a CFR of at least 90% against a bacterial population were considered optimum.

Microbiological analyses
Microbiological data used in the pharmacodynamic model were obtained from the microbiology laboratory database of Hospital São Paulo, a 750-bed university-affi liated tertiary hospital belonging to the Federal University of São Paulo (São Paulo, Brazil).
The data aggregated in the present study were generated from bacterial isolates obtained from blood cultures harvested from hospitalized patients between 2008 and 2010.Identifi cation and antimicrobial susceptibility testing of bacterial species were conducted either by conventional biochemical methodologies or by using the automated BD Phoenix system (Becton Dickinson, Franklin Lakes, NJ, USA).Antimicrobial susceptibility testing was interpreted according to the 2012 Clinical Laboratory Standards Institute (CLSI) guidelines.
The 614 bacterial isolates analyzed in this study included 194 A. baumannii, 192 Klebsiella pneumoniae, 89 P. aeruginosa, and 70 Escherichia coli strains, as well as 69 strains of Enterobacter spp.The MICs necessary to inhibit 50% and 90% (MIC 50 and MIC 90 susceptibility rates) of the growth of each isolate are listed in Table 1 and Table 2. Of the antibiotics tested, piperacillin/tazobactam exhibited the lowest susceptibility rates against E. coli, K. pneumoniae, A. baumannii, and P. aeruginosa; however, this compound was also associated with the highest susceptibility rates against Enterobacter spp.Notably, none of the antimicrobials exhibited susceptibility rates against P. aeruginosa greater than 60%, while the susceptibility rates for each of the antibiotics tested against A. baumannii was greater than 27%.
The PTA for each dosing regimen of cefepime, ceftazidime, imipenem/cilastatin, meropenem, and piperacillin/tazobactam are depicted in Figure 1A and Figure 1E, respectively.
The CFRs for the various simulated antimicrobial dosing regimens are summarized in Table 3. Piperacillin/tazobactam failed to achieve a CFR > 90% for any of the isolates examined, regardless of the MIC and dosing regimens tested.Meanwhile, the highest CFRs for cefepime and piperacillin/tazobactam were obtained in E. coli.Indeed, nearly optimal results were obtained with the 2.0g q8h (3h infusion) and 4.5g q6h

DISCUSSION
Nosocomial infections caused by Gram-negative bacteria are notoriously diffi cult to treat owing to their limited antimicrobial susceptibility and the frequent emergence of resistant mutants during therapy (16) .In the hospital setting, β-lactam antibiotics are commonly used to treat these infections and are the most commonly prescribed antimicrobial class for treatment of Gramnegative bacterial infections (17) .
Studies have shown that the PK/PD parameter that best predicts the effi cacy of β-lactam antibiotics is the length of time for which the free drug remains above the MIC for the infecting microorganism (ƒT > MIC) between dosing intervals.Thus, adjusting the dosage and infusion time can infl uence the effectiveness of these antimicrobials (18) (19) .
Indeed, in this study, we observed that higher doses and prolonged infusions resulted in greater CFRs for all of the β-lactams tested.Specifi cally, prolonged infusion of high-dose meropenem achieved the target CFR for E. coli, Enterobacter spp, and K. pneumoniae (>90%).Furthermore, a 3-h infusion of 2g meropenem approached the target CFR for P. aeruginosa (89.7%).Likewise, regimens comprising prolonged infusion of high-dose ceftazidime also reached the CFR for E. coli, Enterobacter spp, and K. pneumoniae, but not for P. aeruginosa or A. baumannii.Lastly, regardless of the dosage used, prolonged infusion of imipenem-cilastatin achieved the pharmacodynamic target for E. coli and Enterobacteriaceae only, while no cefepime or piperacillin-tazobactam regimen achieved the CFR for any of the organisms tested.Similar results have been reported in previous studies (1) (11) (15) .Koomanachai et al. (1) simulated optimized standard dosing of antimicrobials used in US hospitals for treatment of E. coli, K. pneumoniae, A. baumannii, and P. aeruginosa infections by using MIC data from a surveillance program (1) .These authors demonstrated that prolonged infusion of high-dose β-lactams was associated with increased exposure and enhanced pharmacodynamic results against Enterobacteriaceae and Gram-negative nonfermenters.While the CFR results obtained by this group were superior to those observed in the present study, these fi ndings were likely because the percentage of β-lactam susceptibility was signifi cantly lower in the current study.Furthermore, in an MCS using standard and optimized doses of β-lactam agents against Gram-negative bacteria isolated in 2009 in Canada, Keel et al. found that ideal CFRs were obtained against species from the Enterobacteriaceae family (Enterobacter cloacae, E. coli, and K. pneumoniae) by using standard dosing.Conversely, standard dosing yielded suboptimal CFRs against P. aeruginosa and Acinetobacter spp., suggesting that treatment of these organisms requires high-dose antimicrobial therapies (11) .In our study, only prolonged infusion (2g q8h for 3h each) of high-dose meropenem achieved the ideal PK/PD target for treating P. aeruginosa.The signifi cant impact of increased dosing and prolonged infusions on the effectiveness of carbapenems against P. aeruginosa has been reported in previous studies (1) (11) (20) .
It is noteworthy that P. aeruginosa is an important cause of BSI in humans, and is associated with high mortality rates, ranging from 18-61% (21) .Therefore, we emphasize the importance of empirical use of aggressive β-lactam doses for patients with risk factors for infection by P. aeruginosa or other multidrug-resistant bacteria.
None of the antimicrobial drug regimens tested yielded an ideal CFR against A. baumannii, which is consistent with the observed mechanisms of resistance of this organism, whereby the presence of several β-lactamases typically results in  1, the MIC 50 values of each of the β-lactams tested for A. baumannii were several dilutions higher than those for other microorganisms.Indeed, the therapeutic options for multidrug-resistant A. baumannii are currently limited, and infections by these organisms typically result in poor clinical outcomes.Therefore, drug combination therapy has been suggested for treatment of such infections (20) .In a recent study, Housman et al. conducted an in vitro pharmacodynamics humansimulated exposures of ampicillin/sulbactam, doripenem, and tigecycline alone and combination against multidrug-resistant A. baumannii (22) .Their results demonstrate that therapeutic regimens comprised of combinations of aggressive doses of antimicrobials provide enhanced activity against A. baumannii.Specifi cally, the authors concluded that when polymyxins are not an option, aggressive doses of ampicillin/sulbactam combined with doripenem or tigecycline may be suitable for treating infections caused by sulbactam-susceptible A. baumannii.
Notably, in this study, antimicrobial therapies achieved lower CFRs than those achieved with the same drug regimens in North America for the majority of the strains examined (1) (11) .It is also noteworthy that the percentage of antimicrobial susceptibility directly infl uences the ability of the agent to reach the target (CRF > 90%), and resistance rates in our study were high.Our results are consistent with those reported in previous surveillance studies comparing antimicrobial resistance rates in developing countries (ex.Brazil vs. the US), including the resistance rates of P. aeruginosa to imipenem (47.2% vs. 23.0%,respectively), of K. pneumoniae to ceftazidime (76.3% vs. 27.1%,respectively), and of E. coli to ceftazidime (66.7% vs. 8.1%, respectively) (23) .
Interestingly, ceftazidime reached greater CFRs than cefepime against Enterobacteriaceae in our study.These fi ndings were likely due to the reduced usage of ceftazidime in our hospital in recent years.This factor might also explain the increased rates of susceptibility to ceftazidime among Gram-negative rod-shaped organisms observed in our facility.However, another possible explanation is that there is a higher prevalence of cefotaximase (CTX-M)-producing Enterobacteriaceae strains in Brazil than in other countries (24) .
The present study aimed at providing assessment data that might infl uence pharmacodynamic clinical guidelines for the selection of appropriate empirical antibiotic therapies for bacteremia.To the best of our knowledge, only two other studies have attempted to utilize PK/PD analyses to determine optimal dosing regimens for treatment of nosocomial infections caused by Gram-negative bacteria in Brazil (25) (26) .In addition to the use of such MCS data, however, it is still important to consider the individual MIC values for the infecting bacteria to enable the selection of an adequate treatment.
There are several limitations to this study.First, while previous studies have shown differences in the susceptibility rates of organisms isolated from ICU and non-ICU infections, we were unable to separate the infections based on the location of origin in these hospitals.Second, it was not possible to determine the MICs for each isolate by the broth microdilution method.Instead, we employed an automated MIC approach, which might have underestimated or overestimated the susceptibility rates.Third, while our results were obtained from a large university-affi liated hospital, they cannot be extrapolated to other hospitals.
In summary, the results of our study reinforce that prolonged infusion of high-dose β-lactam antimicrobials comprises the most effective treatment of BSIs caused by pathogens of the family Enterobacteriaceae and by nonfermenting rod-shaped bacteria.

TABLE 1 -Percent susceptibility and minimum inhibitory concentration distributions for various antimicrobials simulated against Enterobacteriaceae (Escherichia coli, Enterobacter spp., and Klebsiella pneumoniae) isolated in a Brazilian hospital between 2008 and 2010.
%S: percent susceptibility; MIC: minimum inhibitory concentration.

TABLE 2 -Percent susceptibility and minimum inhibitory concentration distributions for various antimicrobials simulated against nonfermenting Gram-negative bacilli (Acinetobacter baumannii and Pseudomonas aeruginosa) isolated in a Brazilian hospital between 2008 and 2010. Species/antimicrobial MIC (µg/mL) Percentage Percentage of isolates susceptible at MIC (µg/mL)
%S: percent susceptibility; MIC: minimum inhibitory concentration.

TABLE 3 -Comparison of the CFR for antimicrobial regimens tested against Escherichia coli, Enterobacter spp., Klebsiella pneumoniae, Acinetobacter baumannii, and Pseudomonas aeruginosa.
CFR: cumulative fraction of response; E.: Escherichia; K.: Klebsiella; A.: Acinetobacter; P.: Pseudomonas; CI: continuous infusion.Additionally, prolonged infusion of high-dose meropenem achieved the highest CFRs for E. coli, Enterobacter spp., and K. pneumoniae, and approached an ideal CFR for P. aeruginosa (89.7%).Conversely, no antimicrobial regimen achieved an ideal CFR against A. baumannii.Overall, of the carbapenems, meropenem yielded better results than imipenem for all isolates tested.Rev Soc

Bras Med Trop 48(5):539-545, Sep-Oct, 2015
Furtado GH et al. -PK/PD target attainment of intravenous β-lactam regimens increases in the MIC values of each antimicrobial beyond what is clinically attainable with safe dosing regimens.As demonstrated in Table