Pharmacodynamic evaluation of biapenem in peritoneal fluid using population pharmacokinetic modelling and Monte Carlo simulation

International Journal of Antimicrobial Agents 32 (2008) 339–343

Short communication

Pharmacodynamic evaluation of biapenem in peritoneal fluid using population pharmacokinetic modelling and Monte Carlo simulation Kazuro Ikawa a,∗ , Norifumi Morikawa a , Kayo Ikeda a , Hiroki Ohge b , Taijiro Sueda b a

b

Department of Clinical Pharmacotherapy, Graduate School of Biomedical Sciences, Hiroshima University, 1-2-3 Kasumi, Minami-ku, Hiroshima 734-8551, Japan Department of Surgery, Graduate School of Biomedical Sciences, Hiroshima University, 1-2-3 Kasumi, Minami-ku, Hiroshima 734-8551, Japan Received 8 March 2008; accepted 22 March 2008

Abstract This study evaluated the pharmacodynamics of biapenem in peritoneal fluid (PF). Biapenem (300 or 600 mg) was administered via a 0.5-h infusion to 19 patients before abdominal surgery. Venous blood and PF samples were obtained after 0.5, 1, 2, 3, 4, 5 and 6 h. Drug concentration data (108 plasma samples and 105 PF samples) were analysed using population pharmacokinetic modelling. A three-compartment model fits the data, with creatinine clearance (CLCr ) as the most significant covariate: CL (L/h) = 0.036 × CLCr + 4.88, V1 (L) = 6.95, Q2 (L/h) = 2.05, V2 (L) = 3.47, Q3 (L/h) = 13.7 and V3 (L) = 5.91, where CL is the clearance, Q2 and Q3 are the intercompartmental clearances, and V1, V2 and V3 are the volumes of distribution of the central, peripheral and peritoneal compartments, respectively. A Monte Carlo simulation using the pharmacokinetic model showed the probabilities of attaining the bactericidal exposure target (30% of the time above the minimum inhibitory concentration (T > MIC)) in PF were greater than or equal to those in plasma. In the cases of CLCr = 90 and 60 mL/min, the sitespecific pharmacodynamic-derived breakpoints (the highest MIC values at which the probabilities of target attainment in PF were ≥90%) were 2 ␮g/mL for 300 mg every 12 h, 4 ␮g/mL for biapenem 300 mg every 8 h (q8h) and 8 ␮g/mL for 600 mg q8h. Thus, these results should support the clinical use of biapenem as a treatment for intra-abdominal infections and facilitate the design of the dosing regimen. © 2008 Elsevier B.V. and the International Society of Chemotherapy. All rights reserved. Keywords: Carbapenem; Probability of target attainment; Breakpoint

1. Introduction Intra-abdominal infections (IAIs) are associated with considerable morbidity and mortality. Early use of appropriate empirical antibacterial therapy is vital to ensure the highest probability of a favourable microbiological and clinical outcome. It is particularly important to select broad-spectrum agents that penetrate sufficiently into the abdominal cavity and to optimise their dosing regimens to achieve adequate bactericidal concentrations in the cavity. Biapenem is a broad-spectrum agent used for the empirical treatment of IAIs as well as for antibacterial prophylaxis in abdominal surgery [1]. However, little is known about how well this drug penetrates into an intra-abdominal site, thus ∗

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

attaining a pharmacodynamic (PD) target associated with near maximal bacterial effects at the site. Like other carbapenems, biapenem exhibits a time-dependent killing, therefore its antibacterial effects correlate with the exposure time that the drug concentration remains above the minimum inhibitory concentration (MIC) for the bacterium (T > MIC) [2]. The T > MIC targets required for bacteriostatic and bactericidal (3 log killing) effects in infection models have been reported to be 17% and 30%, respectively [3]. This study evaluated the peritoneal pharmacodynamics of biapenem to support the clinical use of biapenem as a treatment for IAIs and to rationalise empirical dosing regimens. A previously reported population pharmacokinetic modelling and a Monte Carlo simulation approach [4,5] were applied to estimate the probabilities that biapenem regimens attained the bactericidal exposure target in plasma and peritoneal fluid (PF).

0924-8579/$ – see front matter © 2008 Elsevier B.V. and the International Society of Chemotherapy. All rights reserved. doi:10.1016/j.ijantimicag.2008.03.011

340

K. Ikawa et al. / International Journal of Antimicrobial Agents 32 (2008) 339–343

2. Materials and methods 2.1. Study patients Patients who underwent abdominal surgery for the relief of inflammatory bowel disease were chosen as the study subjects because they have a sufficient amount of PF for sampling. The inclusion criteria were: patients of both sexes over 20 years of age; patients who elected to receive antibacterial prophylaxis for a laparotomy; patients willing and able to provide written informed consent; and patients who were not pregnant or who had no history of allergy to ␤lactams. 2.2. Biapenem administration and sample collection Biapenem (300 or 600 mg) was administered via a 0.5-h infusion before abdominal surgery and was then given postoperatively at 8 h intervals for 2 days. Venous blood and PF samples were obtained at the end of the first infusion and 1, 2, 3, 4, 5 and 6 h thereafter. The exudate fluid in the abdominal cavity was manually collected with a syringe during surgery and was obtained post-operatively through an intra-abdominal drain. The plasma and supernatant PF were removed after centrifugation and were then stabilised with an equal volume of 1 M 3-morpholino-propanesulfonic acid buffer (pH 7.0) and stored at −40 ◦ C. 2.3. Biapenem assay The concentrations of biapenem in plasma and PF were determined by high-performance liquid chromatography as reported previously [6]. Briefly, plasma and PF samples (400 ␮L) were transferred to an ultrafiltration device and centrifuged. Then, 20 ␮L of the filtered solution was injected onto a chromatograph with a C18 column and an ultraviolet absorbance detector set at 300 nm. A mixture of 0.1 M sodium acetate buffer (pH 4.6) and acetonitrile (197:3) was used as the mobile phase at a flow rate of 1 mL/min. The lower limit of quantification was 0.04 ␮g/mL and the coefficients of variation were within 8% both in plasma and PF. 2.4. Population pharmacokinetic modelling Population pharmacokinetic modelling was performed using the NONMEM program (version VI; ICON Development Solutions, Ellicott, MD). A standard three-compartment model (1, central; 2, peripheral; 3, peritoneal) was chosen as the basic pharmacokinetic model because it described the current data set better than a two-compartment model (1, central; 2, peritoneal). The fixed-effects parameters were clearance (CL), the volume of distribution of the central compartment (V1), the intercompartmental (central–peripheral) clearance (Q2), the volume of distribution of the peripheral compartment (V2), the intercompartmental (central–peritoneal)

clearance (Q3) and the volume of distribution of the peritoneal compartment (V3). The interindividual variability was modelled exponentially: θi = θ × exp(η)

(1)

where θ i is the fixed-effects parameter for the ith subject, θ is the mean value of the fixed-effects parameter in the population and η is a random interindividual variable that is normally distributed with a mean of 0 and a variance of ω2 . The residual (intraindividual) variability was modelled using a proportional error model: Cobs,ij = Cpred,ij × (1 + ε)

(2)

where Cobs,ij and Cpred,ij denote the jth observed and predicted concentrations for the ith subject and ε is a random intraindividual error that is normally distributed with a mean of 0 and a variance of σ 2 . In the modelling, the first-order conditional estimation (FOCE) method was used. The influence of patient characteristics (sex, age, body weight, blood urea nitrogen, serum creatinine and creatinine clearance (CLCr )) on the individual pharmacokinetic parameters obtained from the basic model was graphically explored. The covariates showing a correlation with the pharmacokinetic parameters were introduced into the basic model (Eq. (3)) one at a time, as expressed by Eqs. (4) and (5): P = θk

(3)

P = θk × (Cov) + θk+1

(4)

P = θk × (Cov)θk+1

(5)

where P is the population pharmacokinetic parameter, θ k and θ k+1 are the mean values of the kth and (k + 1)th fixed-effects parameters in the population and Cov is the value of the examined covariate. The significance of the influence of the examined covariate was evaluated by changes in −2 log likelihood (the minimum value of the objective function, OBJ). An OBJ decrease of more than 3.84 from the basic model (P < 0.05, χ2 test) was considered to be statistically significant during the forward inclusion process. The full model was built by incorporating the significant covariates, and the final model was developed by a backward deletion method. The covariates in the full model were excluded from the model one at a time, and an OBJ increase of more than 7.88 from the full model (P < 0.005, χ2 test) was considered to be statistically significant. 2.5. Monte Carlo simulation A 10 000-subject Monte Carlo simulation was performed to estimate T > MIC in plasma and PF for each biapenem regimen (0.5-h infusion)–MIC combination, where the total drug concentration was employed because it is assumed to be clinically equivalent to the free concentration owing to negligible protein binding (2.3% [7]; 3.7% [8]). The following process was repeated from the 1st to 10 000th subject

K. Ikawa et al. / International Journal of Antimicrobial Agents 32 (2008) 339–343

341

using Crystal Ball 2000 (Decisioneering, Denver, CO). A set of fixed-effects parameters (CL, V1, Q2, V2, Q3 and V3) was generated randomly according to each mean estimate (θ) and the interindividual variance (ω) of the population pharmacokinetic model. The drug concentrations in plasma and PF versus time curves (24 h after the start of the regimen) were simulated using the set of fixed-effects parameters. The time point at which the drug concentration coincided with a twofold diluted MIC value (0.063–64 ␮g/mL) was determined and the exposure time that the drug concentration remained at the MIC was finally calculated as the cumulative percentage for a 24-h period [9]. The probability of target attainment (PTA) was determined as the fractions that achieved at least 30% T > MIC (the bactericidal exposure target) of 10 000 estimates. 2.6. Determination of pharmacodynamic (PD)-derived breakpoint For each biapenem regimen, the PD-derived breakpoints in plasma and PF were determined as the highest MIC values at which the PTAs were ≥90% [10].

3. Results Nineteen abdominal surgery patients received infusions of biapenem 300 mg (n = 15) or 600 mg (n = 4). The demographic and pathophysiological parameters were: sex, 10 men and 9 women; mean ± standard deviation (S.D.) age, 50.0 ± 16.7 years; mean ± S.D. body weight, 51.5 ± 8.2 kg; mean ± S.D. blood urea nitrogen, 10.3 ± 4.1 mg/dL; mean ± S.D. serum creatinine, 0.69 ± 0.24 mg/dL; and mean ± S.D. CLCr , 89.9 ± 22.3 mL/min. A total of 108 concentration samples for plasma (Fig. 1a) and 105 samples for PF (Fig. 1b) were used for the population pharmacokinetic modelling. In the basic model, Q2, V2 and V3 were evaluated as a fixed value without any interindividual variability because their η values were <0.00001. During the forward inclusion process to build the covariate model, multiplicative incorporation of CLCr into CL caused the largest OBJ change (OBJ, −16.4), although age, body weight and CLCr each also had a significant effect on CL. Because age and body weight each showed a high correlation with CLCr , they were not additionally incorporated into CL to avoid a collinearity effect. However, none of the examined covariates had a significant effect on V1 and Q3. During the backward deletion process, CLCr and the coefficient remained in the model, thus causing a significant OBJ increase. Therefore, the mean values of the final population pharmacokinetic parameters were CL (L/h) = 0.036 × CLCr + 4.88; V1 (L) = 6.95; Q2 (L/h) = 2.05; V2 (L) = 3.47; Q3 (L/h) = 13.7 and V3 (L) = 5.91, and their variabilities were ωCL = 0.151, ωV1 = 0.635, ωQ2 = 0, ωV2 = 0, ωQ3 = 0.409, ωV3 = 0 and ε = 0.183. The standard error for each parameter estimate

Fig. 1. Observed concentrations and simulation curves after 0.5-h infusion of biapenem 300 mg (—䊉—) or 600 mg (- - -- - -) in (a) plasma and (b) peritoneal fluid. The simulation curves are illustrated using the mean fixedeffects parameters in the case of creatinine clearance of 90 mL/min.

was <50%, and good fit regression lines were shown in the observed drug concentration (y) versus individual predicted concentration (x) after the Bayesian step: y = 1.071x + 0.004 (r = 0.991) for plasma and y = 0.992x − 0.223 (r = 0.981) for PF. Based on the final population pharmacokinetic model and the range of the study patients’ data (44.8–135.8 mL/min), three patient populations (CLCr = 120, 90 and 60 mL/min) were supposed. In the mean population (CLCr = 90 mL/min), the simulated biapenem concentrations after a dose of 300 mg reached the maximum of 25.9 ␮g/mL at 0.5 h for plasma (Fig. 1a) and 15.6 ␮g/mL at 0.77 h for PF (Fig. 1b), and remained higher in PF than in plasma at 0.78 h post-dose. The PTAs for every regimen were greater in PF (Fig. 2b) than in plasma (Fig. 2a) and the PTAs increased in the following order: 300 mg every 12 h (q12h) < 300 mg every 8 h (q8h) ≈ 600 mg q12h < 600 mg q8h. The PD-derived breakpoints in PF were greater than or equal to those in plasma and these values varied with the CLCr (Table 1). In the cases of CLCr = 90 mL/min and 60 mL/min, the breakpoints in PF were 2 ␮g/mL for 300 mg q12h, 4 ␮g/mL for 300 mg q8h and 600 mg q12h, and 8 ␮g/mL for 600 mg q8h, respectively.

342

K. Ikawa et al. / International Journal of Antimicrobial Agents 32 (2008) 339–343

Fig. 2. Probability of target attainment (PTA) of 30% T > MIC in (a) plasma and (b) peritoneal fluid using different biapenem regimens: 300 mg every 12 h (q12h) (), 300 mg every 8 h (q8h) (), 600 mg q12h () and 600 mg q8h (). The broken line represents 90% PTA. T > MIC, exposure time that the drug concentration remains above the minimum inhibitory concentration for the bacterium.

4. Discussion Earlier animal experiments demonstrated that biapenem widely penetrates into a wide range of body tissues and fluids [1]; however, in humans little is known about the rate and extent of the drug’s penetration into the abdominal cavity. This study monitored biapenem concentrations in plasma and PF from abdominal surgery patients and demonstrated that intravenous biapenem penetrated into PF rapidly and extensively. The population pharmacokinetic modelling showed that the maximum concentration for PF was more than one-

half of the value for plasma, that the difference in the time to the maximum for the two sites was short, and thereafter the concentrations remained higher in PF than in plasma (Fig. 1). The observation that θ Q3 was much greater than θ CL and θ Q2 also indicated a high peritoneal penetrability. This pharmacokinetic property of biapenem enabled the T > MIC in PF to be greater than or equal to that in plasma. The population pharmacokinetic modelling also demonstrated CLCr to be the most significant covariate that affected the peritoneal pharmacokinetics. However, the effect of CLCr on CL was relatively small, since the mean CL estimates for CLCr = 120, 90 and 60 mL/min were 9.20, 8.12 and 7.04 L/h, respectively. These results are consistent with earlier findings that showed biapenem to be eliminated renally with only a 24h urinary recovery of 46.7–53.1% in elderly healthy subjects [11]. By incorporating the developed pharmacokinetic model into a Monte Carlo simulation, this study estimated the bactericidal PTAs in PF and determined the PD-derived breakpoints, which varied with the CLCr of the patient population. There is no breakpoint for biapenem defined by the Clinical and Laboratory Standards Institute, and the breakpoints defined by the Japanese Society of Chemotherapy are limited to pulmonary infections and sepsis [12]. These are the first breakpoints of biapenem determined for IAIs, particularly based on the pharmacodynamics in the interstitial fluid at the action site but not in plasma. The in vitro activities (MIC90 ) of biapenem against common bacteria in IAIs are 0.12 ␮g/mL for Escherichia coli, 0.5 ␮g/mL for Klebsiella pneumoniae, 0.25 ␮g/mL for Enterobacter cloacae and 16 ␮g/mL for Pseudomonas aeruginosa, although the clinical isolates were collected from various fluids such as blood, urine and sputum in the latest nationwide surveillance [13]. Meanwhile, the PD-derived breakpoint values should be weighted for CLCr = 90 and 60 mL/min more than for CLCr = 120 mL/min, because infected patients often demonstrate renal dysfunction. Therefore, the results (Fig. 2; Table 1) suggest that empirical regimens of 300 mg q12h or 300 mg q8h can be sufficiently bactericidal unless there is a strong suspicion of P. aeruginosa or a bacterium with a MIC > 4 ␮g/mL. From a PD viewpoint, 300 mg q8h (900 mg per day regimen) can compete with 600 mg q12h (1200 mg per day regimen).

Table 1 Pharmacodynamic (PD)-derived breakpoints for biapenem regimens in plasma and peritoneal fluid (PF) of patient populations with various degrees of serum creatinine clearance (CLCr ) Biapenem regimen (0.5-h infusion)

PD-derived breakpoint (␮g/mL) CLCr = 120 mL/min

300 mg q12h 300 mg q8h 600 mg q12h 600 mg q8h q12h, every 12 h; q8h, every 8 h.

CLCr = 90 mL/min

CLCr = 60 mL/min

Plasma

PF

Plasma

PF

Plasma

PF

1 2 2 4

1 2 2 4

1 2 2 4

2 4 4 8

2 4 4 8

2 4 4 8

K. Ikawa et al. / International Journal of Antimicrobial Agents 32 (2008) 339–343

There are some potential limitations in this study. The drug concentrations in PF may be underestimated because inflammatory bowel disease is associated with a greater volume of PF than other diseases. A longer T > MIC may be required for the bactericidal effects of biapenem in humans because 30% T > MIC comes from experiments on animals, although it is considered to be the best target value currently available. Therefore, the PD estimates in the PF of the study patients should provide very useful information on the dosing strategy for biapenem, but they do not derive the most optimal dosing regimen for IAIs. In conclusion, intravenous biapenem penetrated into the PF of abdominal surgery patients both rapidly and extensively. The bactericidal PTAs in PF were greater than or equal to those in plasma and these values varied with the CLCr of the patient population. These PD profiles of biapenem in PF therefore support the clinical use of this drug as a treatment for IAIs. The site-specific PD-derived breakpoints for biapenem regimens are the first to have been proposed and they should facilitate the design of effective regimens for the treatment of IAIs. Funding: No funding sources. Competing interests: None declared. Ethical approval: The Ethics Committee of Hiroshima University Hospital, Japan.

[4]

[5]

[6]

[7]

[8] [9]

[10]

[11]

[12]

References [1] Perry CM, Ibbotson T. Biapenem. Drugs 2002;62:2221–34. [2] Zhanel GG, Wiebe R, Dilay L, Thomson K, Rubinstein E, Hoban DJ, et al. Comparative review of the carbapenems. Drugs 2007;67:1027–52. [3] Takata T, Aizawa K, Shimizu A, Sakakibara S, Watabe H, Totsuka K. Optimization of dose and dose regimen of biapenem based on

[13]

343

pharmacokinetic and pharmacodynamic analysis. J Infect Chemother 2004;10:76–85. Ikawa K, Morikawa N, Hayato S, Ikeda K, Ohge H, Sueda T. Pharmacokinetic and pharmacodynamic profiling of cefepime in plasma and peritoneal fluid of abdominal surgery patients. Int J Antimicrob Agents 2007;30:270–3. Ikawa K, Morikawa N, Matsuda S, Ikeda K, Ohge H, Takesue Y, et al. Peritoneal penetration and pharmacodynamic exposure of intravenous cefozopran in abdominal surgery patients. Int J Antimicrob Agents 2007;30:352–5. Ikeda K, Ikawa K, Ikeda A, Nishikawa Y, Morikawa N. A simple and rapid determination of biapenem in plasma by high-performance liquid chromatography. J Chromatogr B: Anal Technol Biomed Life Sci 2006;844:148–52. Niwa T, Nakamura A, Kato T, Kutsuna T, Katou K, Morita H, et al. Pharmacokinetic study of pleural fluid penetration of carbapenem antibiotic agents in chemical pleurisy. Respir Med 2006;100:324–31. Tarao F, Miura T, Saito A, Sato K. Pharmacokinetic study of biapenem. Jpn J Chemother 1996;44:769–75 [in Japanese]. 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. DeRyke CA, Kuti JL, Nicolau DP. Reevaluation of current susceptibility breakpoints for Gram-negative rods based on pharmacodynamic assessment. Diagn Microbiol Infect Dis 2007;58: 337–44. Kozawa O, Uematsu T, Matsuno H, Niwa M, Takiguchi Y, Matsumoto S, et al. Pharmacokinetics and safety of a new parenteral carbapenem antibiotic, biapenem (L-627), in elderly subjects. Antimicrob Agents Chemother 1998;42:1433–6. Saito A, Inamatsu T, Okada J, Oguri T, Kanno H, Kusano N, et al. Clinical breakpoints in pulmonary infections and sepsis: new antimicrobial agents and supplemental information for some agents already released. J Infect Chemother 1999;5:223–6. Yamaguchi K, Murakami M, Takahashi A, Ishii Y, Iwata M, Itoh K, et al. Nationwide surveillance of parenteral antibiotics containing meropenem activities against clinically isolated strains in 2004. Jpn J Antibiot 2005;58:655–89 [in Japanese].