Optimisation of imipenem regimens in patients with impaired renal function by pharmacokinetic-pharmacodynamic target attainment analysis of plasma and urinary concentration data

International Journal of Antimicrobial Agents 40 (2012) 427–433

Contents lists available at SciVerse ScienceDirect

International Journal of Antimicrobial Agents journal homepage: http://www.elsevier.com/locate/ijantimicag

Optimisation of imipenem regimens in patients with impaired renal function by pharmacokinetic-pharmacodynamic target attainment analysis of plasma and urinary concentration data Kenichi Yoshizawa a , Kazuro Ikawa a,∗ , Kayo Ikeda a , Hiromi Kumon b , Hiroki Ohge c , Norifumi Morikawa a a

Department of Clinical Pharmacotherapy, Hiroshima University, 1-2-3 Kasumi, Minami-ku, Hiroshima 734-8551, Japan Department of Urology, Okayama University, 2-5-1 Shikata-cho, Kita-ku, Okayama 700-8558, Japan c Department of Infectious Diseases, Hiroshima University Hospital, 1-2-3 Kasumi, Minami-ku, Hiroshima 734-8551, Japan b

a r t i c l e

i n f o

Article history: Received 17 April 2012 Accepted 25 June 2012 Keywords: Carbapenem Creatinine clearance Probability of target attainment Breakpoint

a b s t r a c t In this study, a pharmacokinetic-pharmacodynamic (PK-PD) target attainment analysis of imipenem (IPM) in patients with impaired renal function was conducted. IPM (500 mg) was administered via a 0.5-h or 1-h infusion to 27 patients with varying renal function. A population PK model was developed by simultaneously fitting plasma and urinary concentration data. A two-compartment model adequately described IPM pharmacokinetics, and creatinine clearance (CLCr ) was identified as the most significant covariate. A PK-PD simulation predicted the probabilities of attaining the target in plasma [40% of the time above the minimum inhibitory concentration (MIC)] and defined the PK-PD breakpoints (the highest MICs at which the probabilities were ≥90%). In a patient with a CLCr of 90 mL/min, prolongation of infusion time (from 0.5 h to 1.5 h) increased the PK-PD breakpoint from 1 ␮g/mL to 2 ␮g/mL with a 500 mg dose every 8 h (q8h) and from 2 ␮g/mL to 4 ␮g/mL with a 500 mg dose every 6 h (q6h). Meanwhile, in a patient with a CLCr of 20 mL/min, the PK-PD breakpoints for both 0.5-h and 1.5-h infusions were 1 ␮g/mL with a 250 mg dose every 12 h (q12h), 2 ␮g/mL with a 250 mg dose q8h and a 500 mg dose q12h, and 4 ␮g/mL with a 250 mg dose q6h. These results indicate that a shorter dosing interval is beneficial in patients with impaired renal function as it results in greater PK-PD breakpoints and a reduction in excessive maximum plasma concentrations. These results help to optimise IPM regimens, particularly in patients with impaired renal function. © 2012 Elsevier B.V. and the International Society of Chemotherapy. All rights reserved.

1. Introduction Imipenem (IPM), a leading compound of the carbapenem antibiotics, possesses a broad antibacterial spectrum against Grampositive and Gram-negative bacteria, with a high degree of stability against ␤-lactamases (both penicillinase and cephalosporinase). IPM is used to treat complicated urinary tract infections, complicated intra-abdominal infections and severe pneumonia. Its pharmacokinetic (PK) profile is characterised by low plasma protein binding (8.7%), predominantly renal excretion in healthy subjects (72%) and a relatively fast elimination half-life (ca. 1 h) [1,2]. The antibacterial effects of IPM are dependent on the time above the minimum inhibitory concentration (T > MIC) against microorganisms within the dosing interval [3]. Therefore, the PK and pharmacodynamic (PD) properties of IPM determine the

∗ Corresponding author. Tel.: +81 82 257 5296; fax: +81 82 257 5320. E-mail address: [email protected] (K. Ikawa).

optimal dosing regimen, especially in special patient populations, e.g. patients with severe underlying disease and renal impairment as well as elderly patients. With regard to drug-related adverse reactions, such as hypersensitivity, hepatic toxicity and neurotoxicity, seizures are a clinically important toxicity of IPM compared with the other carbapenems, i.e. doripenem and meropenem [4]. It is likely that inhibition of ␥-aminobutyric acid (GABA) receptors with IPM results in the depression of GABA-mediated inhibitory neurotransmission [5,6]. In previous clinical trials, various seizure frequencies ranging from 0.4% (4/997) to 10.3% (3/29) have been reported [7,8]. Risk factors associated with IPM-induced seizures are renal impairment, seizure history, central nervous system (CNS) disorders, excessive drug doses, and infections with Pseudomonas aeruginosa [9]. Whilst a reduction in the dose of IPM is obviously required in patients with impaired renal function, the T > MIC in plasma should be maintained to attain maximum bactericidal activity. Thus, well-balanced dosing regimens are necessary for patients with impaired

0924-8579/$ – see front matter © 2012 Elsevier B.V. and the International Society of Chemotherapy. All rights reserved. http://dx.doi.org/10.1016/j.ijantimicag.2012.06.011

428

K. Yoshizawa et al. / International Journal of Antimicrobial Agents 40 (2012) 427–433

renal function having a significant risk factor of IPM-induced seizures. Previously, we constructed population PK models to characterise the intraperitoneal pharmacokinetics of IPM in patients undergoing abdominal surgery and to determine the dosing regimens for intra-abdominal infections [10,11]. Renal function was not integrated into the population PK models as a significant covariate, as drug concentration data were obtained from patients with normal renal function. Population PK models reflecting renal function make it possible to optimise appropriate IPM dosing regimens for patients with impaired renal function. The aim of this study was (i) to develop a population PK model by simultaneous fitting of plasma and urine concentration data and (ii) to estimate the probability of attaining the target for IPM, particularly in patients with impaired renal function. 2. Materials and methods 2.1. Study patients IPM PK data were obtained from the Department of Urology, Okayama University Hospital (Okayama, Japan) and the Department of Surgery, Hiroshima University Hospital (Hiroshima, Japan). This study was approved by the ethics committees of the two study institutes and was conducted in compliance with the Declaration of Helsinki. All subjects provided written informed consent. 2.2. Imipenem administration and sample collection IPM (500 mg), containing 500 mg of cilastatin sodium, was administered via a 0.5-h or 1-h infusion to patients with varying renal function. Venous blood samples were obtained at 0.5, 1, 1.5, 2, 2.5, 3.5, 4, 4.5, 5, 5.5, 6, 6.5, 8 and 12 h after beginning the infusion, and urine samples were obtained with both their volumes and collection intervals recorded (0–2, 2–4, 4–6, 6–8, 8–10 and 10–12 h). Blood and urine samples were immediately collected into polypropylene tubes that were kept on ice and were centrifuged at 3000 × g for 10 min at 4 ◦ C. The supernatants were stabilised with 1 mol/L of 3-morpholino-propanesulphonic acid buffer (pH 7.0) and stored at −80 ◦ C until assay. 2.3. Imipenem assay Plasma and urine concentrations of IPM were measured using high-performance liquid chromatography as previously reported [10]. Briefly, following pre-treatment with an ultrafiltration device, sample aliquots were injected onto a chromatograph with a ␮Bondasphere C18 column (Waters Corp., Milford, MA), and IPM was detected at an ultraviolet absorbance of 300 nm. The lower limit of quantification for IPM was 0.05 ␮g/mL. 2.4. Population pharmacokinetic modelling Population PK modelling was performed with NONMEM 7 (ICON Development Solutions, Dublin, Ireland) and model building was implemented using the subroutine ADVAN3 with the first-order conditional estimation method with interaction. A twocompartment model was fitted to plasma and urine data and was selected as the basic PK model based on preliminary analysis (Fig. 1). Fixed-effects parameters were renal clearance (CLr ), volume of distribution in the central compartment (V1 ), non-renal clearance (CLnr ), intercompartmental clearance (Q) and volume of distribution in the peripheral compartment (V2 ). Interpatient variability on each fixed-effects parameter was modelled using an exponential error structure (log-normal distribution). Two residual

V2

Peripheral X2 Q

Rin V1

Central X1

CLr

Urine X3

CLnr

Fig. 1. Schematic of the basic pharmacokinetic model. X1 , X2 and X3 , amounts of imipenem (mg) in the central, peripheral and urine compartments, respectively; Rin , drug infusion rate (mg/h); CLr and CLnr , renal and non-renal clearance (L/h); Q, intercompartmental clearance (L/h); V1 and V2 , volumes of distribution in the central and peripheral compartments (L).

error models, i.e. proportional, and combined additive and proportional, were also evaluated. The base PK model was selected based on a decrease in the minimum value of NONMEM objective function (−2 log likelihood) and precision of parameter and error estimates. Patient factors, i.e. age, body weight, blood urea nitrogen, serum creatinine, creatinine clearance (CLCr ) and sex, were assessed as potential covariates. First, these potential covariates on CLr , V1 and CLnr were screened via graphical correlation analysis. Then, the continuous covariates were tested for potential relationships with CLr , V1 and CLnr using linear, exponential and power models, whereas sex was tested for potential relationships using a categorical model. Covariates were considered as ‘potentially significant’ when the addition of individual patient factors into the base PK model resulted in a decrease of ≥6.635 points in the NONMEM objective function (2 distribution, P < 0.01). After the full PK model incorporating all of the possible covariates was built, the significance of the potential covariates was evaluated using a backward elimination method. Each covariate was individually removed from the full PK model. The covariates that resulted in an increase of ≥10.828 points in the NONMEM objective function (2 distribution, P < 0.001) were retained in the final PK model. The reduction process was repeated until all remaining covariates were deemed statistically significant. The predictive ability of the final PK model and the randomness of the individual weighted residuals were assessed by goodness-of-fit plots. The final PK model was validated by parameter sensitivity and leverage analyses. Parameter sensitivity analysis was performed by fixing the parameter of interest to specified percentage changes of the population estimate and allowing NONMEM to estimate all other parameters. The curve of the NONMEM objective function value versus parameter value was fitted using a polynomial regression to obtain a 95% confidence interval (CI). Leverage analysis was performed by randomly omitting 10% of the patients from the data set, with each patient being eliminated only once. The final PK model was run using the remaining 90% of the data. The parameter estimates from all runs were compared with the 95% CIs calculated in the parameter sensitivity analysis.

2.5. Microbiological data The MIC distribution data were obtained from the European Committee on Antimicrobial Susceptibility Testing (EUCAST), which provides information on a wide range of microorganisms and antimicrobial agents [12]. Five common types of pathogenic bacteria were selected, specifically Escherichia coli, P. aeruginosa, Haemophilus influenzae, Staphylococcus aureus and Streptococcus pneumoniae. The fractions of isolates of the species at each MIC category are listed in Table 1.

K. Yoshizawa et al. / International Journal of Antimicrobial Agents 40 (2012) 427–433

429

Table 1 Minimum inhibitory concentration (MIC) distributions against clinical isolates of Escherichia coli, Pseudomonas aeruginosa, Haemophilus influenzae, Staphylococcus aureus and Streptococcus pneumoniae. Species

E. coli P. aeruginosa H. influenzae S. aureus S. pneumoniae

Total no.

16 667 63 999 3828 5837 1643

Fraction of isolates at each MIC category (␮g/mL) ≤0.063

0.125

0.25

0.5

1

2

4

8

16

≥32

0.172 0.002 0.056 0.864 0.892

0.454 0.005 0.083 0.052 0.062

0.287 0.018 0.269 0.030 0.036

0.063 0.131 0.354 0.021 0.006

0.013 0.386 0.150 0.011 0.001

0.007 0.199 0.062 0.014 0.002

0.002 0.068 0.016 0.003 0.001

0.001 0.066 0.006 0.002 0

0.001 0.107 0.003 0.001 0

0.001 0.019 0.001 0.002 0

MIC50 (␮g/mL)

MIC90 (␮g/mL)

0.125 1 0.5 0.063 0.063

0.25 16 1 0.125 0.125

MIC50 , MIC for cumulative 50% of isolates; MIC90 , MIC for cumulative 90% of isolates.

2.6. Probability of target attainment (PTA) analysis and pharmacokinetic-pharmacodynamic simulation A PK-PD simulation was conducted using the final PK model to predict the distribution (median, and 5th and 95th percentiles) of plasma concentrations and cumulative urinary excretions of IPM and to estimate the PTA profile for the potential dosing regimens of IPM (0.5-h versus 1.5-h infusions). The following process was repeated 10 000 times with S-PLUS 7.0 (TIBCO Software Inc., Palo Alto, CA). A set of fixed-effects parameters was generated randomly according to each population estimate (), interpatient variability (ω) and residual variability (ε) of the final PK model. The steady-state unbound IPM concentration–time curve was simulated using the fixed-effects parameters, where a value of 10% plasma protein binding was employed [10]. The time point at which the free drug concentration coincided with a two-fold diluted MIC (0.063–32 ␮g/mL) was determined, and the time for which the free drug concentration remained above the MIC (fT > MIC) was finally calculated as a cumulative percentage over a 24-h period [13]. The PTA (%) was determined as a fraction that achieved ≥40% fT > MIC in 10 000 simulation replicates. For each potential dosing regimen of IPM, the PK-PD breakpoint was determined at the highest MIC at which the PTA in plasma is ≥90%. For each bacterium, the PTA at a specific MIC was multiplied by the fraction of the population of bacterium in each MIC category, and the sum of the individual products was defined as the expected PTA for the population (%) [13].

resulted in a 35.566 point drop in the NONMEM objective function and a reduction in interpatient variability in CLr from 81.6% to 34.1% compared with the base PK model. The goodness of fit for the final PK model indicated that there was an agreement between the observed concentrations and predicted values, and a random distribution of weighted residual values. Parameter sensitivity analysis demonstrated that all parameters were estimated with adequate precision. In addition, leverage analysis showed that all parameter estimates from the patient subsets were within the 95% CIs calculated in the parameter sensitivity analysis. Fig. 2 illustrates the individual plasma IPM concentrations and cumulative urinary excretions of the unchanged form from the initiation of IPM infusion. Plasma IPM concentrations declined in a bi-exponential manner after cessation of IPM infusion, and urinary excretion of IPM was completed by ca. 6 h post infusion. The distribution of predicted plasma concentrations and cumulative urinary excretions (median and 5–95th predicted interval) were overlaid on observed data. The majority of observed data were within the 5th and 95th percentile of the simulated distribution, indicating that the model predicted the plasma concentration versus time observations and urinary excretions of IPM in patients with varying degrees of renal function. IPM is primarily excreted via glomerular filtration, and urinary excretion is dependent on renal function [14]. In the present study, the estimated cumulative urinary excretion of IPM in patients with a CLCr of 20 mL/min decreased by ca. 60% compared with patients with a CLCr of 120 mL/min (Fig. 3). Consistent with the predominant role of glomerular filtration in IPM clearance, the final PK model predicted that IPM renal

3. Results Of the 27 Japanese patients (22 male and 5 female) enrolled in this study, 10 and 17 patients were hospitalised in the Departments of Surgery and Urology, respectively. The median (range) of age, blood urine nitrogen, body weight, CLCr and serum creatinine were 60 years (26–87 years), 12.0 mg/dL (5.0–22.5 mg/dL), 58 kg (40–82 kg), 54.1 mL/min (8.7–130 mL/min) and 1.05 mg/dL (0.50–4.43 mg/dL), respectively. A total of 172 plasma and 81 urine samples were used for population PK modelling. The plasma and urinary concentration data were best described by a two-compartment model with interpatient variability in CLr , V1 and CLnr . The typical population estimates of CLr , V1 , CLnr , Q and V2 were 4.09 L/h, 11.4 L, 4.25 L/h, 3.02 L/h and 3.51 L, respectively, in the base PK model. Incorporation of CLCr , serum creatinine, age and body weight into CLr resulted in a decrease in the NONMEM objective function of ≥6.635. However, to avoid a collinearity effect between CLCr and these potential covariates, serum creatinine, age and body weight were not retained in the full PK model. Sex was identified as a potential significant covariate on V1 during the forward selection process. Removal of CLCr from the full PK model resulted in a significant increase in the NONMEM objective function (, +33.32); however, sex was not included in the final PK model (, +5.992). Table 2 presents the parameter estimates for the final PK model. Inclusion of CLCr in the final PK model

Table 2 Estimates of population pharmacokinetic parameters for imipenem. Parameter

Population estimate

Fixed-effects parameters CLr (L/h) =  1 × [CLCr (mL/min)/54.1]2 1 4.18 0.960 2 CLnr (L/h) =  3 3 3.78 V1 (L) =  4 4 11.4 Q (L/h) =  5 5 3.18 V2 (L) =  6 6 3.56 Interpatient variability interaction term 0.0190 CLr and V1 Interpatient variability 0.341 ωCLr 0.294 ωCLnr 0.189 ωV1 Residual variability 0.0342 εadditive (␮g/mL) 25.0 εproportional (%)

Standard error (%)

7.25 12.8 9.84 6.33 30.6 16.8 108 30.3 68.0 54.6 63.6 11.2

CLr , renal clearance; CLCr , creatinine clearance; CLnr , non-renal clearance; V1 and V2 , volumes of distribution in the central and peripheral compartments; Q, intercompartmental (central–peripheral) clearance.

100

(a)

Cumulative urinary excretion of imipenem (% of dose)

K. Yoshizawa et al. / International Journal of Antimicrobial Agents 40 (2012) 427–433

Plasma concentration of imipenem (µg/mL)

430

100

10

1

0.1 0.05 0

2

4

6 Time (h)

8

10

12

(b)

80 60 40 20 0 0

2

4

6 Time (h)

8

10

12

Fig. 2. (a) Observed plasma imipenem (IPM) concentrations and (b) cumulative urinary excretion after a 0.5-h or 1-h infusion of IPM (500 mg) in 27 patients. Closed circles represent the observed plasma and urine data. The solid line and shaded regions represent the median and the 5th and 95th percentile of the predicted intervals, respectively, calculated from 10 000 simulation replicates.

100

100

80

80

60

60

40

40

20

20

0

0 0

20 40 60 80 100 Creatinine clearance (mL/min)

Predicted renal clearance/total clearance ( ,%)

Observed cumulative urinary excretion of imipenem (

clearance decreases with decreasing CLCr , but there was no distinct trend between non-renal clearance and CLCr . Using the final PK model, three typical patients with a CLCr of 90, 60 and 20 mL/min were tested. The terminal elimination halflives of IPM were prolonged (1.3, 1.5 and 2.2 h, respectively) and the area under the plasma concentration–time curve increased (47.2, 59.5 and 92.8 ␮g h/mL, respectively) with decreasing renal function in the patients. The PTAs for IPM were predicted at a specific MIC (Fig. 4). The PTA increased in the following order: a 250 mg dose of IPM every 12 h (q12h) < a 250 mg dose every 8 h (q8h) and a 500 mg dose q12h < a 250 mg dose every 6 h (q6h) in a typical patient with a CLCr of 20 mL/min. A shorter dosing interval achieved higher PK-PD breakpoints as well as total daily doses. In a typical patient with a CLCr of 90 mL/min, a 1.5-h infusion increased the PK-PD breakpoints from 1 ␮g/mL to 2 ␮g/mL with a 500 mg dose q8h and from 2 ␮g/mL to 4 ␮g/mL with a 500 mg dose q6h compared with a 0.5-h infusion (Table 3). However, a prolonged infusion did not improve the PKPD breakpoints in typical patients with a CLCr of 20 mL/min and 60 mL/min. Table 4 presents the PTAs against recent clinically isolated strains, specifically E. coli, P. aeruginosa, H. influenzae, S. aureus and S. pneumoniae. In a typical patient with a CLCr of 90 mL/min, each dosing regimen achieved a PTA of ≥90% against E. coli [MIC for cumulative 90% of isolates (MIC90 ) = 0.25 ␮g/mL], H. influenzae (MIC90 = 1 ␮g/mL), S. aureus (MIC90 = 0.125 ␮g/mL) and

120

Fig. 3. Relationship between cumulative urinary excretion of imipenem (IPM) and creatinine clearance after a 0.5-h or 1-h infusion of IPM (500 mg). Closed circles represent the observed urinary cumulative excretion. The solid line and shaded regions represent the median and the 5th and 95th percentile of the predicted intervals of the ratio of renal clearance to total clearance, respectively, calculated from 10 000 simulation replicates.

S. pneumoniae (MIC90 = 0.125 ␮g/mL). In a typical patient with a CLCr of 20 mL/min, the tested dosing regimen also achieved ≥90% PTA against E. coli, H. influenzae, S. aureus and S. pneumoniae. None of the dosing regimens exhibited a sufficient bactericidal activity against P. aeruginosa, except for the 1000-mg dose q6h via a 1.5-h infusion in a patient with a CLCr of 90 mL/min, likely due to the higher MIC value (MIC90 = 16 ␮g/mL).

4. Discussion IPM is excreted via glomerular filtration and, in part, via tubular secretion, where it is degraded by dehydropeptidase I in the brush border of the proximal renal tubules [15]. Co-administration with cilastatin prevents the degradation of IPM and competitively inhibits the secretion of IPM in the proximal tubules, which in turn increases the urinary recovery of IPM [14]. Overall tubular secretion accounts for 20% of renal excretion, and IPM clearance correlates with glomerular filtration. Considering the mechanisms by which IPM is excreted, renal function appears to be clinically important in precisely estimating drug concentrations and developing appropriate dosing regimens in patients with impaired renal function. Although several population PK models have been previously constructed in conjunction with plasma creatinine levels and glomerular filtration rate, renal clearance of IPM had not been evaluated [16,17]. By incorporating urinary excretion of its unchanged form, the current population PK model can estimate the effects of renal function on total clearance of IPM and can be applied in a PK-PD simulation to optimise the dosing regimens in patients with impaired renal function. Sakka et al. described the plasma concentration data of IPM by a two-compartment model in critically ill patients with nosocomial pneumonia [18]. Age, body weight, height and body surface area were retained as significant covariates in their final PK model, but not CLCr . This may be because the subjects evaluated had normal renal function. The current study demonstrated that the two-compartment model (Fig. 1) adequately described the plasma and urine data (Figs. 2 and 3) and the results of the population PK modelling were consistent with previous findings. In a typical patient with a CLCr of 122 mL/min, our estimated total clearance [CLr (9.12 L/h) + CLnr (3.78 L/h) = 12.9 L/h; Table 2] and V1 (11.4 L) were similar to those previously reported (total clearance = 12.3 L/h and V1 = 12.2 L). Although age and body weight were identified as significant covariates, CLCr was a more direct and statistically significant predictor for describing the pharmacokinetics of IPM and a more suitable covariate for developing the dosing regimens of IPM in patients with impaired renal function.

K. Yoshizawa et al. / International Journal of Antimicrobial Agents 40 (2012) 427–433

(a)

(b)

100

100

Probability of 40% fT > MIC attainment (%)

Probability of 40% fT > MIC attainment (%)

431

80 60 500 mg q8h 0.5-h infusion 500 mg q6h 0.5-h infusion 1000 mg q8h 0.5-h infusion 1000 mg q6h 0.5-h infusion 500 mg q8h 1.5-h infusion 500 mg q6h 1.5-h infusion 1000 mg q8h 1.5-h infusion 1000 mg q6h 1.5-h infusion

40 20 0

80 60 250 mg q6h 0.5-h infusion 500 mg q8h 0.5-h infusion 500 mg q6h 0.5-h infusion 750 mg q8h 0.5-h infusion 250 mg q6h 1.5-h infusion 500 mg q8h 1.5-h infusion 500 mg q6h 1.5-h infusion 750 mg q8h 1.5-h infusion

40 20 0

0.063 0.125 0.25

0.5

1

2

4

8

16

32

0.063 0.125 0.25

0.5

MIC (µg/mL)

1

2

4

8

16

32

MIC (µg/mL)

(c) Probability of 40% fT > MIC attainment (%)

100 80 60 250 mg q12h 0.5-h infusion 250 mg q8h 0.5-h infusion 250 mg q6h 0.5-h infusion 500 mg q12h 0.5-h infusion 250 mg q12h 1.5-h infusion 250 mg q8h 1.5-h infusion 250 mg q6h 1.5-h infusion 500 mg q12h 1.5-h infusion

40 20 0

0.063 0.125 0.25

0.5

1

2

4

8

16

32

MIC (µg/mL) Fig. 4. Probability of target attainment (PTA) of ≥40% fT > MIC in three typical patients with creatinine clearances of (a) 90 mL/min, (b) 60 mL/min and (c) 20 mL/min using different dosing regimens of imipenem [every 12 h (q12h), every 8 h (q8h) and every 6 h (q6h)]. The dotted line represents 90% PTA. fT > MIC, time for which the free drug concentration remained above the minimum inhibitory concentration.

In this study, differences in the PK-PD target attainments against various bacteria appeared to be due to their varying and specific susceptibilities to IPM. The minimum dosing regimens of 250 mg q12h and 500 mg q8h achieved ≥90% PTA against E. coli, H. influenzae, S. aureus and S. pneumoniae (MIC90 ≤ 1 ␮g/mL; Table 1) in typical patients with a CLCr of 20 mL/min and 90 mL/min (Table 4). However, none of the dosing regimens, i.e. shorter dosing interval and/or prolonged infusion time, exceeded the PTA of ≥90% against P. aeruginosa (MIC90 = 16 ␮g/mL; Table 1). Thus, when P. aeruginosa is identified as a causative microorganism and exhibits MIC values greater than the PK-PD breakpoints (Table 3), other appropriate antibiotics should be chosen.

Recent studies demonstrated that a prolonged infusion of IPM improved the PTAs against various microorganisms compared with shorter infusion times, i.e. 0.5-h versus 3-h infusions [19,20]. In contrast, the current results (Table 3) indicated that the effects of infusion time on PK-PD breakpoints depended on CLCr , and a prolonged infusion did not improve the PK-PD breakpoints in typical patients with a CLCr of 20 mL/min and 60 mL/min. These inconsistent findings may be explained by the patient demographics. In the abovementioned studies [19,20], healthy subjects and patients with ventilator-associated pneumonia were recruited and their CLCr was 91.1 ± 8.2 mL/min and ≥60 mL/min, whereas the patients enrolled in the current study had CLCr ranging from 8.7 mL/min to 130 mL/min. Given that IPM is a time-dependent killing antibiotic

Table 3 Pharmacokinetic-pharmacodynamic (PK-PD) breakpoints for the dosing regimens of imipenem (0.5-h and 1.5-h infusions) in patient populations with varying renal function as determined by creatinine clearance (CLCr ). Imipenem regimen

250 mg q12h (500 mg/day) 250 mg q8h (750 mg/day) 250 mg q6h (1000 mg/day) 500 mg q12h (1000 mg/day) 500 mg q8h (1500 mg/day) 500 mg q6h (2000 mg/day) 750 mg q8h (2250 mg/day) 1000 mg q8h (3000 mg/day) 1000 mg q6h (4000 mg/day)

PK-PD breakpoint (␮g/mL)a,b CLCr = 90 mL/min

CLCr = 60 mL/min

CLCr = 20 mL/min

NT/NT NT/NT NT/NT NT/NT 1/2 2/4 NT/NT 2/4 4/8

NT/NT NT/NT 2/2 NT/NT 2/2 4/4 2/4 NT/NT NT/NT

1/1 2/2 4/4 2/2 NT/NT NT/NT NT/NT NT/NT NT/NT

q12h, every 12 h; q8h, every 8 h; q6h, every 6 h; NT, not tested. a Defined as the highest minimum inhibitory concentration (MIC) at which the probability of 40% fT > MIC (time for which the free drug concentration remained above the MIC) attainment in plasma was ≥90%. b 0.5-h/1.5-h infusions.

432

K. Yoshizawa et al. / International Journal of Antimicrobial Agents 40 (2012) 427–433

Table 4 Expected population probabilities of 40% fT > MIC attainment against Escherichia coli, Pseudomonas aeruginosa, Haemophilus influenzae, Staphylococcus aureus and Streptococcus pneumoniae in patients with varying renal function as determined by creatinine clearance (CLCr ) using different dosing regimens of imipenem (0.5-h and 1.5-h infusions). Imipenem regimen

CLCr = 20 mL/min 250 mg q12h (500 mg/day) 250 mg q8h (750 mg/day) 250 mg q6h (1000 mg/day) 500 mg q12h (1000 mg/day) CLCr = 60 mL/min 250 mg q6h (1000 mg/day) 500 mg q8h (1500 mg/day) 500 mg q6h (2000 mg/day) 750 mg q8h (2250 mg/day) CLCr = 90 mL/min 500 mg q8h (1500 mg/day) 500 mg q6h (2000 mg/day) 1000 mg q8h (3000 mg/day) 1000 mg q6h (4000 mg/day)

Probability of 40% fT > MIC attainment (%)a E. coli

P. aeruginosa

H. influenzae

S. aureus

S. pneumoniae

99.4/99.5 99.7/99.8 99.8/99.9 99.8/99.8

67.3/71.5 78.2/80.5 82.3/84.7 78.8/80.9

95.0/96.4 98.2/98.7 99.0/99.3 98.3/98.7

98.8/99.0 99.4/99.5 99.6/99.7 99.4/99.5

99.9/99.9 100/100 100/100 100/100

99.7/99.8 99.7/99.8 99.8/99.9 99.8/99.9

75.3/79.9 77.8/81.4 83.3/87.2 82.5/85.8

97.5/98.6 98.1/98.8 99.1/99.5 98.9/99.3

99.2/99.5 99.4/99.5 99.6/99.7 99.6/99.7

99.9/100 100/100 100/100 100/100

99.4/99.7 99.8/99.9 99.8/99.9 99.9/99.9

69.8/76.6 78.9/84.0 80.4/84.7 86.8/92.5

95.7/97.7 98.3/99.2 98.5/99.2 99.4/99.7

98.9/99.3 99.4/99.6 99.5/99.7 99.7/99.8

99.9/100 100/100 100/100 100/100

fT > MIC, time for which the free drug concentration remained above the minimum inhibitory concentration; q12h, every 12 h; q8h, every 8 h; q6h, every 6 h. a 0.5-h/1.5-h infusions.

[3], plasma concentrations above the MIC would not be maintained in patients with normal renal function owing to the relatively short elimination half-life of IPM. According to the IPM package insert (PRIMAXIN IV; Merck Sharp & Dohme Corp., Whitehouse Station, NJ), dosing regimens of 250 mg or 500 mg q12h, which are equivalent to 500 mg or 1000 mg q6h in patients with normal renal function, are recommended for patients with a CLCr of 6–20 mL/min/1.73 m2 . However, neurological adverse events such as seizures, confusion and tremor were reported when patients with severe renal impairment (CLCr < 30 mL/min) received 500 mg q12h [21]. Animal studies [22,23] found that intravenous and intracerebroventricular injections of IPM induced seizure discharges, which were accompanied by clonic convulsions in a dose-dependent manner, and that a shorter infusion time at the same dose (which can increase the maximum drug concentration but not the area under the drug concentration–time curve) augmented more the epileptogenic activity of IPM. These findings suggest that higher IPM plasma concentrations may cause seizures, although other risk factors such as seizure history, CNS disorders and P. aeruginosa infections may also be involved. This PK-PD simulation indicated that a 250 mg dose q6h was superior to a 500 mg dose q12h as there were higher PK-PD breakpoints and reductions in excessive maximum plasma concentrations, even though the total daily dose is equivalent to the recommend dose in the package insert. Thus, a shorter dosing interval may be beneficial to patients with impaired renal function. From a safety perspective, a prolonged infusion may also be beneficial to renal dysfunction patients by reducing the risk of IPMinduced seizures owing to decreasing excessive maximum plasma concentrations, even though the prolongation of infusion time did not improve the PK-PD breakpoints (Table 3). Herein, we proposed optimised dosing regimens of IPM for patients with impaired renal function based on PK-PD simulation. However, there are several limitations that need to be addressed. (i) When healthy male subjects received 14 C-labelled IPM or 14 Clabelled cilastatin together with an equal amount of unlabelled counter-component, almost all of the radioactivity was recovered in the urine and consisted of the open lactam metabolite of IPM and the N-acetyl metabolite of cilastatin in addition to the unchanged form. Furthermore, faecal excretion accounted for <1% of the total radioactivity [24]. The elimination half-life of cilastatin is ca. 8 times longer in patients with severely impaired renal function than those with normal renal function [14]. Thus, cilastatin and the metabolites of IPM and/or cilastatin may accumulate after multiple doses in patients with severely impaired renal function and

may increase the incidence of adverse events, despite the fact that optimal doses of IPM were adjusted with the PK-PD simulation. (ii) The PK-PD breakpoints and the PTAs were estimated based on unbound plasma concentrations of IPM. Meanwhile, DahyotFizelier et al. showed that unbound peritoneal fluid concentrations of IPM were lower than the corresponding unbound plasma concentrations [25]. Thus, our proposed regimens may not always achieve their PK-PD breakpoints at a specific site of infection. It is important to pay attention to drug concentrations at infection sites to ensure optimisation of IPM regimens. (iii) The treatment period was not discussed sufficiently. Many reports have previously proposed various dosing regimens for IPM based on PK-PD modelling and simulation that focused on MICs [17–19]. However, they also did not discuss the optimum treatment period for IPM. (iv) The safety and efficacy of the proposed dosing regimens in patients with impaired renal function should be confirmed in future clinical studies. In conclusion, this study developed a more accurate population PK model of IPM, which simultaneously fitted plasma and urine concentration data, and estimated the PTA for IPM, particularly in patients with impaired renal function. This PK-PD simulation indicated that a prolonged infusion improved the PK-PD breakpoints in a patient with a CLCr of 90 mL/min and that a shorter dosing interval was useful in maintaining plasma IPM concentrations above MIC in a patient with a CLCr of 20 mL/min. The developed population PK model helps to optimise the dosing regimens for IPM, particularly in patients with impaired renal function. However, further studies are required to confirm the clinical implications of these findings and proposed dosing regimens. Funding: No funding sources. Competing interests: None declared. Ethical approval: This study was approved by the Ethics Committees of Okayama University (Okayama, Japan) and Hiroshima University (Hiroshima, Japan) and was conducted in compliance with the Declaration of Helsinki. All subjects provided written informed consent.

References [1] Dreetz M, Hamacher J, Eller J, Borner K, Koeppe P, Schaberg T, et al. Serum bactericidal activities and comparative pharmacokinetics of meropenem and imipenem–cilastatin. Antimicrob Agents Chemother 1996;40:105–9. [2] Norrby SR, Alestig K, Björnegård B, Burman LA, Ferber F, Huber JL, et al. Urinary recovery of N-formimidoyl thienamycin (MK0787) as affected by coadministration of N-formimidoyl thienamycin dehydropeptidase inhibitors. Antimicrob Agents Chemother 1983;23:300–7.

K. Yoshizawa et al. / International Journal of Antimicrobial Agents 40 (2012) 427–433 [3] Flückiger U, Segessenmann C, Gerber AU. Integration of pharmacokinetics and pharmacodynamics of imipenem in a human-adapted mouse model. Antimicrob Agents Chemother 1991;35:1905–10. [4] Miller AD, Ball AM, Bookstaver PB, Dornblaser EK, Bennett CL. Epileptogenic potential of carbapenem agents: mechanism of action, seizure rates, and clinical considerations. Pharmacotherapy 2011;31:408–23. [5] Fujimoto M, Munakata M, Akaike N. Dual mechanisms of GABAA response inhibition by beta-lactam antibiotics in the pyramidal neurones of the rat cerebral cortex. Br J Pharmacol 1995;116:3014–20. [6] Sunagawa M, Matsumura H, Sumita Y, Nouda H. Structural features resulting in convulsive activity of carbapenem compounds: effect of C-2 side chain. J Antibiot (Tokyo) 1995;48:408–16. [7] Koppel BS, Hauser WA, Politis C, van Duin D, Daras M. Seizures in the critically ill: the role of imipenem. Epilepsia 2001;42:1590–3. [8] Winston DJ, Ho WG, Bruckner DA, Gale RP, Champlin RE. Controlled trials of double ␤-lactam therapy with cefoperazone plus piperacillin in febrile granulocytopenic patients. Am J Med 1988;85:21–30. [9] Calandra G, Lydick E, Carrigan J, Weiss L, Guess H. Factors predisposing to seizures in seriously ill infected patients receiving antibiotics: experience with imipenem/cilastatin. Am J Med 1988;84:911–8. [10] Ikawa K, Morikawa N, Sakamoto K, Ikeda K, Ohge H, Takesue Y, et al. Pharmacokinetics and pharmacodynamic assessment of imipenem in the intraperitoneal fluid of abdominal surgery patients. Chemotherapy 2008;54:131–9. [11] Ikawa K, Morikawa N, Ikeda K, Ohge H, Sueda T. Development of breakpoints of carbapenems for intraabdominal infections based on pharmacokinetics and pharmacodynamics in peritoneal fluid. J Infect Chemother 2008;14:330–2. [12] European Committee on Antimicrobial Susceptibility Testing (EUCAST). MIC distributions. http://www.eucast.org/mic distributions/ [accessed 18.02.12]. [13] Mouton JW, Dudley MN, Cars O, Derendorf H, Drusano GL. Standardization of pharmacokinetic-pharmacodynamic (PK-PD) terminology for anti-infective drugs: an update. J Antimicrob Chemother 2005;55:601–7. [14] Verpooten GA, Verbist L, Buntinx AP, Entwistle LA, Jones KH, De Broe ME. The pharmacokinetics of imipenem (thienamycin–formamidine) and the renal dehydropeptidase inhibitor cilastatin sodium in normal subjects and patients with renal failure. Br J Clin Pharmacol 1984;18:183–93.

433

[15] Fukasawa M, Sumita Y, Harabe ET, Tanio T, Nouda H, Kohzuki T, et al. Stability of meropenem and effect of 1 ␤-methyl substitution on its stability in the presence of renal dehydropeptidase I. Antimicrob Agents Chemother 1992;36: 1577–9. [16] Dailly E, Kergueris MF, Pannier M, Jolliet P, Bourin M. Population pharmacokinetics of imipenem in burn patients. Fundam Clin Pharmacol 2003;17:645–50. [17] Lamoth F, Buclin T, Csajka C, Pascual A, Calandra T, Marchetti O. Reassessment of recommended imipenem doses in febrile neutropenic patients with hematological malignancies. Antimicrob Agents Chemother 2009;53:785–7. [18] Sakka SG, Glauner AK, Bulitta JB, Kinzig-Schippers M, Pfister W, Drusano GL, et al. Population pharmacokinetics and pharmacodynamics of continuous versus short-term infusion of imipenem–cilastatin in critically ill patients in a randomized, controlled trial. Antimicrob Agents Chemother 2007;51:3304–10. [19] Lee LS, Kinzig-Schippers M, Nafziger AN, Ma L, Sörgel F, Jones RN, et al. Comparison of 30-min and 3-h infusion regimens for imipenem/cilastatin and for meropenem evaluated by Monte Carlo simulation. Diagn Microbiol Infect Dis 2010;68:251–8. [20] Jaruratanasirikul S, Sudsai T. Comparison of the pharmacodynamics of imipenem in patients with ventilator-associated pneumonia following administration by 2 or 0.5 h infusion. J Antimicrob Chemother 2009;63:560–3. [21] Campise M. Neurological complication during imipenem/cilastatin therapy in uraemic patients. Nephrol Dial Transplant 1998;13:1895–6. [22] Horiuchi M, Kimura M, Tokumura M, Hasebe N, Arai T, Abe K. Absence of convulsive liability of doripenem, a new carbapenem antibiotic, in comparison with ␤-lactam antibiotics. Toxicology 2006;222:114–24. [23] Dupuis A, Couet W, Paquereau J, Debarre S, Portron A, Jamois C, et al. Pharmacokinetic–pharmacodynamic modeling of the electroencephalogram effect of imipenem in healthy rats. Antimicrob Agents Chemother 2001;45:1682–7. [24] Norrby SR, Rogers JD, Ferber F, Jones KH, Zacchei AG, Weidner LL, et al. Disposition of radiolabeled imipenem and cilastatin in normal human volunteers. Antimicrob Agents Chemother 1984;26:707–14. [25] Dahyot-Fizelier C, Lefeuvre S, Laksiri L, Marchand S, Sawchuk RJ, Couet W, et al. Kinetics of imipenem distribution into the peritoneal fluid of patients with severe peritonitis studied by microdialysis. Clin Pharmacokinet 2010;49:323–34.