Pharmacodynamic profiling of antimicrobials against Gram-negative respiratory isolates from Canadian hospitals

The objective of this study was to assess the profile of a variety of dosing regimens for common intravenous antibiotics against contemporary Enterobacter cloacae, Escherichia coli, Klebsiella pneumoniae and Pseudomonas aeruginosa isolates collected in Canada during 2009, using pharmacodynamic modelling techniques. Monte Carlo simulation was conducted for standard and/or prolonged infusion regimens of cefepime, ceftazidime, ceftriaxone, ciprofloxacin, doripenem, ertapenem, meropenem and piperacillin/tazobactam. The cumulative fraction of response (CFR) was calculated using bactericidal targets for each regimen against each species. All cefepime, doripenem, ertapenem and meropenem regimens achieved optimal exposures against Enterobacteriaceae, whereas target attainment was organism and dose dependent for the other agents. These results support that the currently recommended antimicrobial dosing regimens generally attain acceptable exposures to achieve the requisite pharmacodynamic targets against the Enterobacteriaceae species; however, they fall short of obtaining optimal bactericidal exposures against P aeruginosa.

Abstract

BACKGROUND:

With diminishing antimicrobial potency, the choice of effective empirical therapy has become more challenging. Thus, the pharmacodynamic evaluation of potential therapies is essential to identify optimal agents, doses and administration strategies.

METHODS:

Monte Carlo simulation was conducted for standard and/or prolonged infusion regimens of cefepime, ceftazidime, ceftriaxone, ciprofloxacin, doripenem, ertapenem, meropenem and piperacillin/tazobactam. Minimum inhibitory concentrations were obtained for Escherichia coli (n=64 respiratory isolates), Enterobacter cloacae (n=53), Klebsiella pneumoniae (n=75) and Pseudomonas aeruginosa (n=273) throughout Canada. The cumulative fraction of response (CFR) was calculated using bactericidal targets for each regimen against each species. A CFR ≥90% was defined as optimal.

RESULTS:

All cefepime, doripenem, ertapenem and meropenem regimens achieved optimal exposures against Enterobacteriaceae, whereas target attainment was organism and dose dependent for the other agents. Prolonged infusion doripenem and meropenem 1 g and 2 g every 8 h, along with standard infusion doripenem and meropenem 2 g every 8 h, were the only regimens to attain optimal exposures against P aeruginosa. Ciprofloxacin had the lowest CFR against P aeruginosa, followed by cefepime. Among the P aeruginosa isolates collected in the intensive care unit (ICU) compared with the wards, differences of 0.5% to 10% were noted in favour of non-ICU isolates for all agents; however, marked differences (10% to 15%) in CFR were observed for ciprofloxacin in favour of ICU isolates.

CONCLUSION:

Standard dosing of cefepime, doripenem, ertapenem and meropenem has a high likelihood of obtaining optimal pharmacodynamic indexes against these Enterobacteriaceae. For P aeruginosa, aggressive treatment with high-dose and/or prolonged infusion regimens are likely required to address the elevated resistance rates of respiratory isolates from Canada.

Gram-negative bacilli are often implicated as causative pathogens in pneumonia (1). Unfortunately, resistance rates among Gram-negative pathogens are rising, and infections due to these resistant strains have been associated with increased mortality, longer hospital stays and higher hospital costs (2). As resistance rates increase, the effective antimicrobial armamentarium against these pathogens dwindles (3), and antimicrobial surveillance becomes vital in determining the role of various antibiotics in the treatment of serious Gram-negative infections. Moreover, it has been found that inadequate empirial therapy is associated with increased patient morbidity and mortality (4). For these reasons, it is prudent that clinicians prescribe antimicrobial regimens that have a high likelihood of achieving microbiological success until susceptibilities are known and de-escalation can occur.

The ability of an antimicrobial dosing regimen to attain bactericidal exposures is often predictive of microbiological response. Although microbiological success does not always equate to clinical success, considering it does not take into account host or process factors, previous pharmacodynamic modelling predictions have shown a correlation between microbiological and clinical success (5,6). As such, Monte Carlo simulation, an interpretation of in vitro potency and pharmacokinetics, has been used to analyze the ability of various antimicrobial dosing regimens to achieve maximal bactericidal pharmacodynamic exposures against target pathogens.

The objective of the current PASSPORT (Probability of target attainment of Antibacterial agents Studied for Susceptibility and Pharmacodynamic Optimization in Regional Trials) was to use pharmacodynamic modelling techniques to assess the profile of a variety of dosing regimens for common intravenous antibiotics against contemporary Enterobacter cloacae, Escherichia coli, Klebsiella pneumoniae and Pseudomonas aeruginosa isolates collected in Canada during 2009.

METHODS

Microbiology and susceptibility testing

A total of 53 E cloacae, 64 E coli, 75 K pneumoniae and 273 P aeruginosa nonduplicate respiratory isolates were collected in the Canadian Ward (CANWARD) surveillance study (www.can-r.ca). Each site submitted unique (one organism per infection site per patient), consecutive, clinically significant isolates during 2009. The minimum inhibitory concentrations (MICs) were assessed using broth microdilution according to the Clinical and Laboratory Standards Institute (CLSI)-defined methodology, and were classified as susceptible, intermediate or resistant according to CLSI interpretive criteria for each antimicrobial, if available. Otherwise, the United States Food and Drug Administration susceptibility breakpoints were used. New breakpoints for the carbapenems against Enterobacteriaceae were approved by the CLSI in 2010. The susceptibility breakpoints were changed to ≤1 μg/mL for doripenem and meropenem, and to ≤0.25 μg/mL for ertapenem. Per cent susceptibility was also calculated and reported for comparison using these new breakpoints.

Antimicrobials

Antimicrobials were included in the pharmacodynamic model if they had comparative MIC data available from CANWARD. The following intravenous regimens were simulated, which included approved regimens as well as off-label high-dose, prolonged and continuous infusion regimens:

• Cefepime 1 g every 6 h and every 8 h (0.5 h and 3 h infusions), 1 g and 2 g every 12 h (0.5 h infusion), and 2 g every 8 h (0.5 h and 3 h infusions).

• Ceftazidime 1 g and 2 g every 8 h (0.5 h and 3 h infusions).

• Ceftriaxone 1 g and 2 g every 24 h (0.5 h infusion).

• Ciprofloxacin 0.4 g every 8 h and every 12 h (1 h infusion).

• Doripenem 0.5 g, 1 g and 2 g every 8 h (1 h and 4 h infusions).

• Ertapenem 1 g every 24 h (0.5 h infusion).

• Meropenem 0.5 g every 6 h (0.5 h and 3 h infusions); 0.5 g, 1 g and 2 g every 8 h (0.5 h and 3 h infusions).

• Piperacillin/tazobactam 3.375 g and 4.5 g every 6 h and every 8 h (0.5 h infusion); 3.375 g every 8 h (4 h infusion); 4.5 g every 6 h (3 h infusion); and 9 g, 13.5 g and 18 g every 24 h (continuous infusion).

Monte Carlo simulation

Steady-state exposures were determined, as previously described (7), for each antimicrobial regimen using serum pharmacokinetic parameters obtained from published population pharmacokinetic studies of infected and/or critically ill adult patients (8–15). The pharmacokinetic parameters used for each antimicrobial are available from a previous PASSPORT publication (16), with the exception of ceftriaxone (7). A 5000-patient Monte Carlo simulation (Crystal Ball 2000, Decisioneering Inc, USA) was conducted for each regimen as previously described (7,16). The volume of the central compartment, total body clearance, k₂₁ and k₁₂ were assumed to follow log-Gaussian distributions, while the unbound fraction was assumed to follow a uniform distribution in which any value within the simulated range had an equal probability of occurring. The Monte Carlo simulation determined the probability of a simulated patient achieving the predefined pharmacodynamic target at a specific MIC dilution. This is referred to as the probability of target attainment (PTA), which was calculated over a range of doubling MICs between 0.008 μg/mL and 256 μg/mL. The pharmacodynamic targets were defined as the percentage of the dosing interval in which the free drug concentration is above the MIC (fT>MIC) – at least 40% for carbapenems, 50% for penicillins and 60% for cephalosporins (17,18). For ciprofloxacin, a ratio of the area under the concentration curve to MIC of at least 125 was the pharmacodynamic target used (19).

PTAs for each antimicrobial regimen were used to determine the cumulative fraction of response (CFR) against the entire bacterial population. The regimen’s simulation-derived PTA was multiplied by the percentage of isolates at each MIC dilution and summed to obtain the CFR. A CFR ≥90% was considered to be optimal against the bacterial population.

RESULTS

The MIC required to inhibit the growth of 50% and 90% of organisms, as well as the susceptibility rates for all respiratory isolates, are shown in Table 1. By using the new carbapenem breakpoints, the per cent susceptibility changed minimally (<2%) for doripenem and meropenem against all three Enterobacteriaceae; however, for ertapenem, a larger change (3% to 10%) in susceptibility was noted. Piperacillin/tazobactam had the highest susceptibility rate (89.0%) against P aeruginosa, followed closely by meropenem (85.3%) and doripenem (82.4%).

TABLE 1.

MIC₅₀, MIC₉₀ and percentage of susceptibility (% S) of respiratory Enterobacter cloacae, Escherichia coli, Klebsiella pneumoniae and Pseudomonas aeruginosa isolates collected in Canada in 2009

Antimicrobial	E cloacae(n=53)		E coli(n=64)		K pneumoniae(n=75)		P aeruginosa(n=273)
	MIC50	% S	MIC50	% S	MIC50	% S	MIC50	MIC90	% S
Cefepime	0.25	100	0.25	93.8	0.25	94.7	4	16	75.5
Ceftazidime	0.5	85.0	0.25	86.0	0.25	94.7	4	32	78.0
Ceftriaxone	0.25	75.5	0.25	82.8	0.25	94.7	32	>64	25.3
Ciprofloxacin	0.06	96.2	0.06	70.3	0.06	90.7	0.5	8	65.2
Doripenem	0.03	98.1 (98.1)	0.03	100 (100)	0.03	100 (100)	0.5	8	82.4
Ertapenem	0.03	98.1 (88.7)	0.03	100 (96.9)	0.03	100 (94.6)	8	32	NT
Meropenem	0.03	100 (98.1)	0.03	100 (100)	0.3	100 (100)	0.5	8	85.3
Piperacillin/tazobactam	2	88.7	2	98.4	4	96.0	4	>64	89.0

All minimum inhibitory concentration (MIC) values required to inhibit the growth of 50% and 90% of organisms (MIC₅₀ and MIC₉₀) are expressed as μg/mL. % S calculated for each carbapenem against E coli, K pneumoniae and E cloacae using new breakpoints are provided in parentheses, where applicable. NT Not tested

A summary of CFRs for all antimicrobial regimens against E cloacae, E coli, K pneumoniae and P aeruginosa are presented in Table 2. Cefepime, doripenem, ertapenem and meropenem at standard doses were able to to achieve optimal exposures against Enterobacteriaceae species; however, there was discordance with regard to which standard dosing regimens obtained optimal CFRs for ceftazidime, ceftriaxone, ciprofloxacin and piperacillin/tazobactam against the three Enterobacteriaceae. The percentage of susceptibility was in agreement with the antimicrobial’s ability to attain optimal exposures.

TABLE 2.

Cumulative fraction of response (CFR) of intravenous antimicrobial regimens against Enterobacter cloacae, Escherichia coli, Klebsiella pneumoniae and Pseudomonas aeruginosa

Antibiotic regimen (infusion duration)	CFR,%
	E cloacae	E coli	K pneumoniae	P aeruginosa
Cefepime
1 g q 12 h*†	95.6	90.7	92	56
2 g q 12 h*†	97.4	93.6	93.9	70.9
1 g q 8 h	98.2	93.8	94.4	68.3
2 g q 8 h*†	99.2	96.3	96	80.7
1 g q 6 h	99.2	95.4	95.5	76
1 g q 8 h (3 h)	99.2	94.9	95.3	72.8
2 g q 8 h (3 h)	99.7	97.1	96.6	84.6
1 g q 6 h (3 h)	99.8	96.3	96	81.3
Ceftazidime
1 g q 8 h	85	86.2	94.7	75
2 g q 8 h*†	87.3	89.1	95.5	85.1
1 g q 8 h (3 h)	85.3	87	94.8	81.1
2 g q 8 h (3 h)	91.8	93.2	97.1	91.5
Ceftriaxone
1 g q 24 h*	73	81.7	94.1	NT
2 g q 24 h*	76.9	84.1	94.7	NT
Ciprofloxacin
0.4 g q 12 h (1 h)	96.2	66.8	88.1	45.8
0.4 g q 8 h (1 h)*†	96.2	67.5	89	52.8
Doripenem
0.5 g q 8 h (1 h)	97.4	99	98.8	79.8
1 g q 8 h (1 h)	98.5	99.6	99.4	86.5
2 g q 8 h (1 h)	99.2	99.8	99.7	91.6
0.5 g q 8 h (4 h)	99.1	100	100	89.4
1 g q 8 h (4 h)	99.7	100	100	94.3
2 g q 8 h (4 h)	99.9	100	100	97.8
Ertapenem
1 g q 24 h	96.2	99.6	99.3	NT
Meropenem
0.5 g q 8 h	99.4	100	100	78
0.5 g q 6 h*	99.7	100	100	82.2
1 g q 8 h*†	99.7	100	100	84.5
2 g q 8 h	99.9	100	100	90.2
0.5 g q 8 h (3 h)	99.9	100	100	85.2
0.5 g q 6 h (3 h)	100	100	100	87.9
1 g q 8 h (3 h)	100	100	100	90.3
2 g q 8 h (3 h)	100	100	100	94.9
Piperacillin/tazobactam
3.375 g q 8 h	80.7	89.2	81.1	67.4
4.5 g q 8 h*	83.4	91.5	85.4	71.7
3.375 g q 6 h	87	94.8	90.3	76.3
4.5 g q 6 h*	89.4	96.1	92.9	79.2
3.375 g q 8 h (4 h)	90.6	98	96	81.9
4.5 g q 6 h (3 h)	94.8	98.4	97.6	85.1
9 g q 24 h (CI)	89.2	97.4	95	80.8
13.5 g q 24 h (CI)	92.4	98.3	96.7	83.1
18 g q 24 h (CI)	94.5	98.4	97.5	84.8

All antibiotics simulated as 0.5 h infusion unless noted after dosing regimen.

^*Recommended by the Association of Medical Microbiology and Infectious Disease Canada guidelines for the treatment of hospital-acquired pneumonia for those with risk factors for multidrug-resistant pathogens, suspected P aeruginosa or severe illness.

^†Recommended by the Infectious Disease Society of America for the treatment of hospital-acquired pneumonia for those with late-onset disease or risk factors for multidrug-resistant pathogens. CI Continuous infusion; NT Not tested; q Every

Against P aeruginosa, increasing the dose and/or prolonging the infusion duration had the greatest predicted impact on the PTA. The only regimens that obtained optimal exposures were high-dose, standard infusion doripenem and meropenem (2 g every 8 h), prolonged infusion of doripenem and meropenem (1 g and 2 g every 8 h), and prolonged infusion ceftazidime (2 g every 8 h). Ciprofloxacin has the lowest predicted CFR (45.8% to 52.8%) of all the tested antimicrobials against this P aeruginosa population. When comparing isolates collected in the intensive care unit (ICU) with those collected outside of the ICU (Table 3), the CFRs were generally in favour of the non-ICU isolates (0.5% to 10%). Interestingly, for ciprofloxacin, there were distinct differences (approximately 10% to 15%) in CFR, which were in favour of isolates collected within the ICU.

TABLE 3.

Comparison of the cumulative fraction of response (CFR) for antibiotic regimens against Pseudomonas aeruginosa respiratory isolates collected in the ICU compared with those collected outside of the ICU

Antibiotic regimen (infusion duration)	CFR,%
	P aeruginosa
	ICU (n=81)	Non-ICU (n=192)
Cefepime
1 g q 12 h*†	56.1	56
2 g q 12 h*†	70.5	71
1 g q 8 h†	68	68.4
2 g q 8 h*†	80.1	81
1 g q 6 h	75.5	76.2
1 g q 8 h (3 h)	72.4	72.9
2 g q 8 h (3 h)	83.9	84.9
1 g q 6 h (3 h)	80.7	81.5
Ceftazidime
1 g q 8 h	69.3	77.4
2 g q 8 h*†	80	87.2
1 g q 8 h (3 h)	75.2	83.6
2 g q 8 h (3 h)	88.3	92.9
Ciprofloxacin
0.4 g q 12 h (1 h)	54.1	42.4
0.4 g q 8 h (1 h)*†	61.2	49.3
Doripenem
0.5 g q 8 h (1 h)	77.3	80.8
1 g q 8 h (1 h)	84.8	87.2
2 g q 8 h (1 h)	90.5	92
0.5 g q 8 h (4 h)	87.5	90.2
1 g q 8 h (4 h)	93.4	94.7
2 g q 8 h (4 h)	97.5	97.9
Meropenem
0.5 g q 8 h	71.8	80.6
0.5 g q 6 h*	76.7	84.6
1 g q 8 h*†	79.3	86.7
2 g q 8 h	86	92
0.5 g q 8 h (3 h)	79.9	87.5
0.5 g q 6 h (3 h)	83.1	89.9
1 g q 8 h (3 h)	85.8	92.1
2 g q 8 h (3 h)	91.6	96.2
Piperacillin/tazobactam
3.375 g q 8 h (0.5 h)	59.2	70.9
4.5 g q 8 h (0.5 h)*	64	75
3.375 g q 6 h (0.5 h)	69	79.5
4.5 g q 6 h (0.5 h)*†	72.5	82.1
3.375 g q 8 h (4 h)	75.2	84.8
4.5 g q 6 h (3 h)	80.1	87.2
9 g q 24 h (CI)	83.6	83.8
13.5 g q 24 h (CI)	76.9	85.7
18 g q 24 h (CI)	79.6	86.9

All antibiotics simulated as 0.5 h infusion unless noted after dosing regimen.

^*Recommended by the Association of Medical Microbiology and Infectious Disease Canada guidelines for the treatment of hospital-acquired pneumonia for those with risk factors for multidrug-resistant pathogens, suspected P aeruginosa or severe illness;

^†Recommended by the Infectious Disease Society of America for the treatment of hospital-acquired pneumonia for those with late-onset disease or risk factors for multidrug-resistant pathogens. CI Continuous infusion; ICU Intensive care unit; q Every

DISCUSSION

Resistance among several Gram-negative pathogens continues to be a growing issue associated with worse clinical outcomes (20,21). With few novel antimicrobials on the horizon (3), pharmacodynamic optimization of currently available agents is essential to maximize efficacy and minimize further development of resistance. The present study evaluated numerous intravenous treatment options from the pharmacodynamic perspective to identify optimal empirical agents and dosing regimens against core Canadian respiratory pathogens. The design of the study predicted microbiological success by calculating the CFR using the pharmacodynamic properties of each included antimicrobial.

Minimal studies have been undertaken to determine the target attainment of antimicrobials against Gram-negative pathogens in Canada (22); however, surveillance studies reporting susceptibility and resistance rates have been published (23,24). If the percentage of susceptibility was in concordance with the CFR of the standard dosing regimens against all three Enterobacteriaceae species tested, susceptibility rates should be relatively accurate at determining whether a regimen is an optimal empirical choice. Although this was true for all three Enterobacteriaceae species, the CFRs for standard dosing regimens against P aeruginosa were typically lower than the reported percentage of susceptibility. Of note, the MIC distributions used in this pharmacodynamic evaluation were obtained nationally from a contemporary collection of E cloacae, E coli, K pneumoniae and P aeruginosa isolates; however, individual hospital or local resistance rates might vary considerably from the national surveillance.

To attain reliable empirical therapy against P aeruginosa or other resistant organisms, adjustments in doses or infusion times can affect the efficacy of the beta-lactams. It has been shown that the percentage of the dosing interval in which the free drug concentration remains above the MIC of the infecting organism (fT>MIC) is the pharmacokinetic/pharmacodynamic parameter that best predicts efficacy (25). To optimize this parameter, and ultimately antimicrobial efficacy of the beta-lactams, prolonged infusion of 3 h to 4 h can be used rather than the standard 0.5 h to 1 h infusion. Although the susceptibility of the organism is often taken into account, the dose, infusion time and the antimicrobial MIC should be considered. While the safety and efficacy of these high doses, in conjunction with prolonged infusion, have not been evaluated in well-controlled clinical trials, data from animal and limited human studies support the safe attainment of in vivo exposures sufficient for these higher MIC organisms (26,27).

By simulating higher doses, prolonged infusions or a combination of both, an increase in predicted CFR was observed for all beta-lactam antibiotics; however, the more susceptible the population, the less of a difference dose and infusion optimization made. Accordingly, the largest impact of dose and infusion optimization was observed against P aeruginosa. Despite improvements in CFR, only ceftazidime (2 g every 8 h as a 3 h infusion), doripenem (2 g every 8 h as a 1 h or 4 h infusion; 1 g every 8 h as a 4 h infusion), and meropenem (2 g every 8 h as a 0.5 h or 3 h infusion; 1 g every 8 h as a 3 h infusion) met the a priori definition of optimal against P aeruginosa. This necessitates the importance of aggressive empirical dosing if the patient has risk factors for P aeruginosa or other multidrug-resistant organisms. Additionally, the CFRs for the P aeruginosa isolates collected in the ICU were generally lower than the isolates collected outside of the ICU, indicating a shift to the right in MIC distributions. This, however, was not true for ciprofloxacin, which had higher CFRs in the ICU population compared with the non-ICU population. While it is not known why the ICU isolates were more susceptible to ciprofloxacin, the CFR results were consistent with the ciprofloxacin MIC distributions for the ICU and non-ICU isolates.

When comparing CFRs with the recommended treatments for hospital-acquired pneumonia, some recurring themes become evident. In general, regardless of whether evaluating using the Infectious Disease Society of America (28) or the Association of Medical Microbiology and Infectious Disease Canada (AMMI) guidelines (1), the recommended antimicrobials and dosing regimens are sufficient to attain reliable coverage, with few caveats against the Enterobacteriaceae species, but not P aeruginosa. Even the higher or more frequent doses recommended by AMMI for severe illness or the most aggressive doses endorsed by the Infectious Disease Society of America fall short of obtaining a CFR ≥90% against P aeruginosa. This reiterates the need for empirically optimizing an antimicrobial’s pharmacodynamics for individuals with risk factors for multidrug-resistant pathogens including pseudomonas. Additionally, although ciprofloxacin is often thought to be an antipseudomonal fluoroquinolone and is recommended as monotherapy by AMMI for patients with mild to moderate hospital-acquired pneumonia at risk for resistant pathogens, it had the lowest CFR even when given three times daily (52.8%). Moreover, ciprofloxacin failed to attain reliable coverage against E coli (0.4 g every 12 h, [66.8%]; 0.4 g every 8 h [67.5%]).

The present pharmacodynamic evaluation was intended to offer guidance to clinicians as they initiate empirical therapy for respiratory infections in Canada. These data emphasize that the currently recommended antimicrobial dosing regimens generally attain acceptable exposures to achieve the requisite pharmacodynamic targets against the Enterobacteriaceae species; however, they fall short of obtaining optimal bactericidal exposures against P aeruginosa. Higher dosages in conjunction with prolonged infusion improved the activity of all beta-lactams in achieving their pharmacodynamic target against all tested organisms; however, the dose optimization made the largest impact on P aeruginosa isolates. This emphasizes the importance of aggressive dosing of appropriate antibiotics when patients are at risk for multi-drug-resistant pathogens including P aeruginosa.