- Email: [email protected]
Comparison of the pharmacokinetics and pharmacodynamic profile of carumonam in cystic fibrosis patients and healthy volunteers
Available online at www.sciencedirect.com
Diagnostic Microbiology and Infectious Disease 65 (2009) 130 – 141 www.elsevier.com/locate/diagmicrobio
Comparison of the pharmacokinetics and pharmacodynamic profile of carumonam in cystic fibrosis patients and healthy volunteers☆ Jürgen B. Bulittaa,1 , Stephen B. Duffullb , Cornelia B. Landersdorfera,1 , Martina Kinziga , Ulrike Holzgrabec , Ulrich Stephana,d , George L. Drusanoe , Fritz Sörgela,d,⁎ a
IBMP—Institute for Biomedical and Pharmaceutical Research, D-90562 Nürnberg-Heroldsberg, Germany b School of Pharmacy, University of Otago, 9054 Dunedin, New Zealand c Institute of Pharmacy and Food Chemistry, University of Würzburg, D-97074 Würzburg, Germany d Department of Pharmacology, University of Duisburg–Essen, D-45122 Essen, Germany e Ordway Research Institute, Albany, NY 12208, USA Received 31 March 2008; accepted 18 June 2009
Abstract Our objectives were to compare the pharmacokinetics (PK) of carumonam, a monobactam, between cystic fibrosis (CF) patients and healthy volunteers and assess its pharmacodynamic profile. We studied 10 adult CF patients and 18 healthy volunteers of similar body size (dose: 2.166 g of carumonam as 15-min intravenous infusion). High performance liquid chromatography with ultraviolet detection (HPLCUV) was used for drug analysis and NONMEM® (ICON, Ellicot City, MD) for population PK and Monte Carlo simulation with targets between ≥20% and 100% free time above MIC (f T N MIC). Unscaled renal clearance was 24% higher in CF patients. Lean body mass and creatinine clearance explained the difference in average clearance and volume of distribution between both subject groups. For a daily dose of 6 g per 70 kg of total body weight, 15-min infusions q8h achieved robust (N90%) probabilities of target attainment (PTAs) (target, 60% f T N MIC) for MICs ≤3 mg/L in CF patients and ≤6 mg/L in healthy volunteers. At the same dose, 4-h infusions q8h achieved robust PTAs up to markedly higher MICs ≤8 to 12 mg/L in CF patients and ≤16 mg/L in healthy volunteers. © 2009 Elsevier Inc. All rights reserved. Keywords: Cystic fibrosis/healthy volunteers; Population pharmacokinetics/pharmacodynamics of carumonam; Monte Carlo simulation; Allometric scaling by lean body mass; PK–PD MIC breakpoints
1. Introduction In Caucasians, cystic fibrosis (CF) is the most common inherited disease. Abnormal chloride transport across the apical membrane of epithelial cells of CF patients causes a high degree of morbidity and mortality. The most serious ☆
This work was presented in part as part of a pharmacokinetic metaanalysis at the 45th Interscience Conference on Antimicrobial Agents and Chemotherapy, 2005 (poster abstract A-12), as part of a pharmacodynamic meta-analysis at the AAPS Annual Meeting, 2006 (poster W4056), and as a chapter in J. Bulitta's PhD thesis. ⁎ Corresponding author. IBMP—Institute for Biomedical and Pharmaceutical Research, D-90562 Nürnberg-Heroldsberg, Germany. Tel.: +49911-518-290; fax: +49-911-518-2920. E-mail address: [email protected] (F. Sörgel). 1 Present address: Department of Pharmaceutical Sciences, State University of New York at Buffalo, NY 14260, USA. 0732-8893/$ – see front matter © 2009 Elsevier Inc. All rights reserved. doi:10.1016/j.diagmicrobio.2009.06.018
hazard is inspissated secretions in the airways. Those secretions host many bacteria that can cause respiratory tract infections. The high morbidity and mortality involved in this pulmonary disease underline the need for optimal antiinfective treatment of CF patients. After initial infection by Haemophilus influenzae or Staphylococcus aureus, many CF patients experience infections by Pseudomonas aeruginosa, methicillin-resistant S. aureus, or Burkholderia cepacia. Therefore, it is important to predict and optimize the probability of successful antibiotic treatment against those pathogens as a function of their MIC distribution in CF patients. Carumonam is an N-sulfonated monobactam antibiotic that is relatively stable against β-lactamases (Imada et al., 1985; Jones et al., 1986; Raimondi et al., 1988; Zhao et al., 2008) and is marketed as Amasulin® by Takeda in Japan. Carumonam is active against H. influenzae (Ikemoto et al.,
J.B. Bulitta et al. / Diagnostic Microbiology and Infectious Disease 65 (2009) 130–141
1998), Klebsiella pneumoniae (Kumamoto et al., 2006), P. aeruginosa (Ikemoto et al., 1996b, 1997; Shinagawa et al., 1996), Escherichia coli (Kumamoto et al., 2006), and Serratia marcescens (Igari et al., 2003; Kumamoto et al., 2003) and might be a treatment option for respiratory or urinary tract infections caused by those pathogens (Yoshida et al., 1986). Typical daily doses range from 1 to 2 g of carumonam split into 2 doses (Sweetman, 2009). Doses of 2 g q8h were well tolerated in healthy volunteers (Berube et al., 1988). We are not aware of any study on the pharmacokinetics (PK) of carumonam in CF patients. For many antibiotics, there are too few controlled clinical trials that systematically compare the PK of antibiotics in CF patients and healthy volunteers (Gottschalk et al., 1988; Prandota, 1987, 1988; Rey et al., 1998; Sorgel et al., 1987; Touw, 1998; Touw et al., 1998). This lack of data may hamper optimal dose selection in CF patients. Many PK studies in CF patients lack a control group of healthy volunteers, which makes a PK comparison of CF patients and healthy volunteers difficult. One difference between CF patients and healthy volunteers is that CF patients are, on average, leaner than healthy volunteers. Body size and body composition have been shown to influence PK (Cheymol, 2000; Green and Duffull, 2004; Morgan and Bray, 1994; Paap and Nahata, 1990; Touw et al., 1998). This requires dose adjustment in some patient groups. Because CF patients are often lean, their lean body mass (LBM) makes up a larger proportion of their total body weight (WT) compared with healthy volunteers (Rochat et al., 1994; Slosman et al., 1992). From a PK point of view, the ideal size descriptor should be able to describe body size and body composition for a large variety of patient groups, for example, for obese, undernourished, and “normal” patients. The ideal size descriptor should also reduce the unexplained (random) between-subject variability (BSV) in PK parameters because this would allow one to achieve target concentrations more precisely in empiric therapy. Population PK can directly use body size, body composition, and renal function to explain the observed variability in PK parameters between patients (e.g., for clearance and volume of distribution). The ability of various body size models to describe the difference in average clearance and volume of distribution and to reduce the unexplained BSV can be directly compared via population PK. There are reports that clearance and volume of distribution can be better described by LBM than by WT in CF patients (Miller and Kornhauser, 1994; Prandota, 1987; Rey et al., 1998; Touw et al., 1994, 1996, 1998). Touw et al. (1994, 1996) proposed to use LBM instead of WT for initial dose selection of tobramycin in CF patients. Although LBM has been proposed to be a superior predictor for dose selection compared with WT or body surface area (Morgan and Bray, 1994), the potential merits of dose selection based on LBM in CF patients still need to be shown (Morgan and Bray, 1994; Rey et al., 1998). Our primary objective was to compare the population PK of carumonam between CF patients and healthy volunteers
131
and to study whether LBM is a more appropriate descriptor of body size and functional capacity than WT. As secondary objective, we assessed the probability of target attainment (PTA) for CF patients and healthy volunteers for various dosage regimens based on PK–pharmacodynamic (PD) targets via Monte Carlo simulation (MCS). We used the time during which the non–protein-bound plasma concentration exceeds the minimal inhibitory concentration (f T N MIC) in combination with PK–PD targets between 20% and 100% f T N MIC. 2. Materials and methods 2.1. Subjects A total of 28 Caucasian volunteers (10 CF patients and 18 healthy volunteers) of both sexes participated in the study after they had given their written informed consent. The subjects' health status was assessed by physical examination, electrocardiography, and laboratory tests including urinalysis and screening for drugs of abuse. Consumption of alcohol in any form or of other medication (including antibiotics) was forbidden from 24 h before carumonam dose until the last sample. After fasting overnight, the subjects received a standardized breakfast at 1 h postdose and a standardized lunch at 4 h postdose. Fluid intake was standardized to 125 mL of mineral water per hour. All volunteers were closely observed by physicians for the occurrence of adverse events during the period of drug administration. The study protocol had been approved by the local ethics committee, and the study was performed according to the revised version of the Declaration of Helsinki. 2.2. Study design and drug administration The study was a single-dose, single-center, open, parallel group study. Each subject received a dose of 2.166 g of carumonam as a 15-min intravenous infusion. All infusions were administered with a BRAUN-Perfusor® (Braun, Melsungen, Germany). The instruments were checked on a daily basis by weighing defined volumes delivered by the Perfusor. 2.3. Blood and urine sampling All blood samples were drawn in vials containing citric acid from a forearm vein via an intravenous catheter contralateral to the one used for drug administration. Blood samples were drawn immediately before start of the infusion (0 min) and at 5, 10, and 15 min poststart of infusion as well as at 5, 15, 20, 30, 45, 60, 75, and 90 min and 2, 3, 4, 5, 6, 8, 10, 12, and 24 h after the end of infusion. After centrifugation, all plasma samples were immediately frozen and stored at −70 °C until analysis. Urine was collected from the start of the infusion until 1 h after the end of infusion as well as at 1 to 2, 2 to 3, 3 to 4, 4 to 6, 6 to 8, 8 to 10, 10 to 12, and 12 to 24 h after end of infusion. The urine samples were
132
J.B. Bulitta et al. / Diagnostic Microbiology and Infectious Disease 65 (2009) 130–141
stored at 4 °C during the collection period. The amount and pH of the urine were measured, and aliquots were frozen and stored at −70 °C until analysis.
The same allometric body size model was applied for LBM with a standard LBM (LBMSTD) of 53 kg.
2.4. Drug analysis in plasma and urine
FSize;V;i =
The concentration of carumonam in plasma and urine was determined by high-performance liquid chromatography. Plasma samples (200 μL) were deproteinized by addition of 400 μL of acetonitrile. This mixture was centrifuged at 10,000 rpm for 5 min, and 1000 μL of dichloromethane were added to the supernatant. Ten microliters of the aqueous phase were injected onto the system. We used a Hypersil ODS 5 (Thermo Fisher Scientific, Dreieich, Germany), 125 × 4 mm column and a 1% acetic acid (glacial)/2% methanol, 0.3% (NH4)2SO4 eluent at pH 2.9. Carumonam was detected at a wavelength of 293 nm. The limit of quantification was 0.2 mg/L for plasma and 7 mg/L for urine. The coefficients of variation for interday precision were between 4.1% and 5.5% for plasma and between 2.6% and 6.0% for urine.
LBMi LBMSTD LBMi 0:75 FSize; CL;i = LBMSTD
ð3Þ ð4Þ
All exponents were set to 1.0 for linear scaling by WT or linear scaling by LBM. 2.5.3. Renal function The glomerular filtration rate of the ith subject (GFRi) was estimated by the Cockcroft and Gault (1976) formula for a nominal 70-kg subject: " agei GFRi = 140 years # mg=dL mL ð5Þ Fsex serum creatinine concentrationi min
2.5. Population PK analysis 2.5.1. Structural model One-, 2-, and 3-compartment disposition models with a time-delimited zero-order input into the central compartment were tested. Competing models were discriminated by their predictive performance assessed via visual predictive checks, NONMEM's objective function (proportional to −2 times the log-likelihood), and residual plots. 2.5.2. Size We studied the following body size models: 1) no body size model, 2) linear scaling by WT, 3) allometric scaling by WT (Anderson and Holford, 2008; Holford, 1996; West et al., 1997, 1999), 4) linear scaling by LBM (Cheymol, 2000; James, 1976), and 5) allometric scaling by LBM. The ability of the body size models to describe the differences in the central tendency of PK parameters was compared between CF patients and healthy volunteers. Furthermore, we compared the ability of the body size models to reduce the unexplained BSV. Our allometric model assumes that volume of distribution scales linearly (allometric exponent 1.0) with body size (i.e., WT or LBM) and that clearance scales slightly less than linearly with body size (allometric exponent, 0.75). The allometric exponent was fixed to 1.0 for all volume terms and fixed to 0.75 for all clearance terms. The FSize,V,i and FSize,CL,i are the fractional changes in volume of distribution and clearance for the ith subject (with WTi) standardized to a weight WTSTD of 70 kg. FSize;V;i =
WTi WTSTD
WTi FSize; CL;i = WTSTD
ð1Þ 0:75
The sex-specific factor (Fsex) is 0.85 for females and 1.0 for males. We scaled the estimated glomerular filtration rate by a standard glomerular filtration rate (GFRSTD) of 120 mL/min per 70 kg. Renal function of the ith subject (RFi) can then be predicted as follows: RFi =
GFRi GFRSTD
ð6Þ
2.5.4. Clearance We assumed a nonrenal and a renal component of clearance (CLNR and CLR, respectively). Renal clearance of carumonam was estimated based on the amount of carumonam excreted unchanged into urine, and renal function, (RF) was used as a covariate on renal clearance. Renal clearance was assumed to be linearly related to RF. 2.5.5. BSV model We estimated the BSV for renal clearance, nonrenal clearance, volume of distribution, and duration of zero-order input by assuming a log-normal distribution for the PK parameters. The log-scale differences (ηBSV) of the individual PK parameters from their population mean are assumed to be normally distributed random variables with mean zero and standard deviation BSV. The BSV was estimated as variance, but we report the square root of the estimate. We expressed these values in the text as a percentage, because this quantity is an approximation to the apparent coefficient of variation of a normal distribution on log scale. The individual PK parameters were calculated as follows (parameters explained below): CLR;i = CLPOP;R FSize;CL;i FCYFCLR RFi expðgBSVCLRi Þ
ð2Þ
ð7Þ
J.B. Bulitta et al. / Diagnostic Microbiology and Infectious Disease 65 (2009) 130–141
CLNR;i = CLPOP;NR FSize;CL;i FCYFCLNR expðgBSVCLNRi Þ ð8Þ CLi = CLR;i + CLNR;i
ð9Þ
The CLi, CLRi, and CLNRi are the individual estimates for total, renal, and nonrenal clearance for the ith subject. The ηBSVCLRi and ηBSVCLNRi are ηBSV of CLR and CLNR for the ith subject, and CLPOP,R and CLPOP,NR are the group estimates for the renal and nonrenal clearance of a healthy volunteer with standard body size (i.e., WTSTD = 70 kg or LBMSTD = 53 kg) and with an RF of 1. The FCYFCLR and FCYFCLNR represent disease-specific scale factors for CF patients that characterize the difference in renal and nonrenal clearance between CF patients and healthy volunteers. A disease-specific scale factor for volume of distribution at steady state (FCYFVSS) was also estimated. 2.5.6. Nonparametric bootstrap Confidence intervals of population PK parameter estimates were determined by nonparametric bootstrap resampling techniques with 1000 replicates for each body size model (Efron and Tibshirani, 1986; Parke et al., 1999). Data from 10 randomly selected CF patients and 18 randomly selected healthy volunteers from the original raw dataset (subjects could be drawn multiple times with replacement) were included in each replicate. The median and nonparametric 90% confidence intervals (5–95% percentile) were calculated from the 1000 estimated PK parameters of our final population PK model. 2.5.7. Observation model and computation The residual unidentified variability was described by a combined additive and proportional error model for plasma concentrations and amounts excreted in urine. We used the first-order conditional estimation method with the interaction estimation option in NONMEM version V release 1.1 (NONMEM Project Group, University of California, San Francisco, CA) (Beal et al., 1999) for all population PK modeling. 2.5.8. Noncompartmental analysis The vast majority of PK studies in CF patients was analyzed by noncompartmental methods (Rey et al., 1998; Touw et al., 1998). For easier comparison of our study with studies in literature, we used WinNonlin™ Professional (version 4.0.1; Pharsight, Mountain View, CA) to calculate noncompartmental PK parameters and descriptive statistics. 2.5.9. Monte Carlo simulation Because carumonam is a β-lactam, we assumed that f T N MIC is the most appropriate PK–PD index. We calculated the PTA for f T N MIC targets of at least 20%, 40%, 60%, 80%, or 100% within the MIC range from 0.25 to 64 mg/L. A protein binding of 23% was used for carumonam, which was shown to be independent of plasma concentration between 25 and 400 mg/L (McNulty et al., 1985).
133
We compared the PTA versus MIC profiles for a daily carumonam dose of 6 g per 70 kg of WT given 1) as continuous infusion, 2) split into three 15-min infusions q8h, or 3) split into three 4-h infusions q8h. Concentration time profiles of 10,000 CF patients and 18,000 healthy volunteers with the same demographic data as the subjects in this study were simulated at steady state in absence of residual error in NONMEM. We used the population mean parameter estimates and parameter variability model of our final population PK model with a block diagonal variance– covariance matrix for MCS; dose, timing of infusion, and protein binding were assumed to have no variability. The f T N MIC values and the PTA were calculated by linear interpolation between simulated data points (19 observations per 8 h) by validated Perl scripts. We derived the PTA for CF patients and healthy volunteers by calculating the fraction of subjects who attained the PK–PD target at each MIC for each target. The PK–PD MIC breakpoint was defined as the highest MIC for which the PTA was at least 90%. The PK–PD MIC breakpoints for the target f T N MIC ≥ 60% are reported in the text because the target for near-maximal bactericidal activity for other β-lactams fall between 40% and 70% f T N MIC (Craig, 1998, 2003; Drusano, 2004). The PTA versus MIC profiles are presented for targets of 20%, 40%, 60%, 80%, and 100% f T N MIC in graphical form to cover a wide range of PK–PD targets. We ran additional simulations based on the individual PK parameter estimates and covariates (WT and LBM) from the 10 CF patients in our study to compare dosing algorithms based on the body size models identified in this analysis as providing the best biologic description of the data (Bulitta et al., 2006). The expected PTA was calculated for Japanese patients with urinary tract infections caused by P. aeruginosa based on their MIC distribution from 2004 (Kumamoto et al., 2006). These MICs ranged from 0.25 to 128 mg/L (MIC50, 4 mg/L; MIC90, 32 mg/L). For this MCS, a daily carumonam dose of 6 g per 70 kg of WT and our final population PK parameter estimates for healthy volunteers were used. To reflect the potentially larger variability in a broader (and potentially more sick) CF patient population, we ran additional MCS with a manually increased BSV in total clearance. The coefficient of variation (%CV) (data are median [10–90% percentile]) of 38 β-lactam datasets in CF patients reviewed by Touw et al. (1998) was 27% (16–53%) for total clearance, 29% (17–53%) for volume of distribution at steady state, and 22% (11–45%) for terminal halflife. The %CV in total clearance from our final estimates was approximately 16%. We subsequently increased the % CV in renal clearance to 34% and 65% to achieve variability in total clearance of approximately 27% and 53%, respectively. The latter 2 estimates are the median and 90% percentile of the distribution of variability in total clearance from the 38 β-lactam datasets in CF patients reviewed by Touw et al. (1998).
134
J.B. Bulitta et al. / Diagnostic Microbiology and Infectious Disease 65 (2009) 130–141
3. Results Our CF patients and healthy volunteers had comparable demographic characteristics (Table 1). The noncompartmental parameter estimates shown in Table 2 are not scaled by any size descriptor. CF patients had a 16% higher unscaled total clearance compared with healthy volunteers, which was primarily caused by a 24% higher unscaled renal clearance. The unscaled volume of distribution at steady state was similar between both subject groups. 3.1. Population PK analysis The visual predictive check indicated a better predictive performance for the 3-compartment model than for the 2compartment model. The 1-compartment model had insufficient predictive performance. Fig. 1 reveals the excellent predictive performance of the 3-compartment model with LBM as size descriptor for plasma concentrations and amounts excreted in urine. The visual predictive check showed that the model was able to well capture the central tendency and variability of the plasma and urine profiles for CF patients and healthy volunteers. NONMEM's objective function favored the 3-compartment model by 164 points relative to the 2-compartment model. Therefore, we chose the 3-compartment model as final structural model. The estimates and bootstrap confidence intervals for our final population PK model based on LBM are shown in Table 3. We studied 1) which body size model best described the difference in average clearance and volume of distribution between CF patients and healthy volunteers and 2) which body size model reduced the unexplained BSV most. The ability of the body size models to describe the difference in average PK parameters between both patient groups is shown in Table 4. The scale factors FCYFCLR, FCYFCLNR, and FCYFVSS are the ratios of group estimates between CF patients and healthy volunteers after accounting for body size and renal function. A value of 1.0 for FCYF means that CF patients and healthy volunteers have the same group estimate Table 1 Demographic data (median [minimum–maximum])
No. of subjects (males/females) Height (cm) Body surface areaa (m2) Age (year) WT (kg) LBMb (kg) Body mass index (kg m−2) Serum creatinine concentration (μmol L−1) Renal functionc (mL min−1) a
CF patients
Healthy volunteers
10 (9/1) 178 (167–183) 1.64 (1.48–1.76) 21 (19–25) 54 (47–61) 47.9 (41.9–52.9) 18.0 (14.5–19.2) 64 (49–80)
18 (12/6) 174 (160–203) 1.73 (1.52–2.46) 25 (21–37) 61 (50–107) 49.7 (40.0–82.2) 20.6 (18.1–26.0) 80 (71–95)
162 (132–193)
115 (97–133)
Calculated by the formula from Mosteller (1987). b Calculated by the formula from Cheymol (2000) and James (1976). c Estimated renal function for a subject with nominal WT of 70 kg (see Materials and methods for details).
Table 2 Unscaled PK parameter estimates from noncompartmental analysis for CF patients and healthy volunteers (median [minimum–maximum])
Total clearance (L/h) Renal clearance (L/h) Nonrenal clearance (L/h) Volume of distribution at steady state (L) Fraction excreted unchanged in urine (%) Peak concentration (mg/L) Terminal half-life (L) Mean residence time (h)
CF patients
Healthy volunteers
7.00 (6.14–8.67) 5.72 (3.08–7.19) 1.55 (0.04–3.64) 12.8 (8.01–15.5)
6.05 (4.94–7.46) 4.62 (2.40–5.93) 1.39 (0.63–2.91) 12.1 (9.65–18.4)
80 (46–99) 207 (174–472) 2.00 (1.43–3.10) 1.88 (1.13–2.19)
76 (45–89) 248 (153–352) 2.02 (1.50–2.61) 2.02 (1.65–2.47)
for the respective PK parameter. A value above (below) 1.0 indicates that CF patients have a higher (lower) estimate for the respective PK parameter compared with healthy volunteers with the same body size and same renal function. It is important to note that the estimate for FCYFCLR was derived for a model that included renal function as a covariate on renal clearance. As shown in Table 1, CF patients had a 41% higher average estimated renal function for a nominal 70-kg subject. The estimates for FCYFCLR (Table 4) were close to 1.0 (range, 0.92–1.07 for the different body size models). Thus, CF patients had both an increased creatinine clearance (renal function) and a higher renal clearance of carumonam compared with healthy volunteers. Linear and allometric scaling by WT predicted total clearance and volume of distribution to be 7% to 11% higher in CF patients relative to healthy volunteers (Table 4A). Linear or allometric scaling by LBM explained the difference in average total clearance and volume of distribution between CF patients and healthy volunteers (range, 1.01–1.04 for FCYFCLT and FCYFVSS) (Table 4A). The bootstrap results were consistent with the results for the original dataset. Total clearance was significantly larger in CF patients compared with healthy volunteers for linear scaling by WT because the 90% confidence interval of FCYFCLT did not include 100% for linear scaling by WT (Table 4B). Table 5 shows the ability of the studied body size models to reduce the unexplained BSV. Linear scaling by WT was used as reference model because doses of CF patients are most commonly calculated on a milligram-per-kilogram WT basis. Allometric scaling by WT reduced the unexplained BSV in renal clearance by 38% (P = 0.027 from bootstrap, 1-sided test), and allometric scaling by LBM reduced the unexplained BSV by 30% (P = 0.135), both compared with linear scaling by WT. As final covariate model, we selected allometric scaling by LBM because this model explained the differences in average clearance between both subject groups and reduced the unexplained BSV similarly to allometric scaling by WT. 3.2. Monte Carlo simulation The PTAs were lower for CF patients compared with healthy volunteers (Fig. 2). The following PK–PD MIC
J.B. Bulitta et al. / Diagnostic Microbiology and Infectious Disease 65 (2009) 130–141
135
Fig. 1. Visual predictive check based on 5000 CF patients and 9000 healthy volunteers for plasma concentrations and amounts excreted unchanged in urine for the 3-compartment model based on LBM (Table 3). The plots show the observations, the 80% prediction interval (10–90% percentile), and the interquartile range (25–75% percentile). Ideally, 50% of the observations should fall inside the interquartile range at each time point, and 80% of the observations should fall inside the 80% prediction interval.
breakpoints represent the highest MIC for which the PTA was at least 90% for the target f T N MIC ≥ 60%. Both subject groups had a PK–PD MIC breakpoint of 16 mg/L for continuous infusion at a daily carumonam dose of 6 g per 70 kg of WT at steady state (Fig. 2E and F). For prolonged (4-h) infusions of 2 g per 70 kg of WT q8h, the PK–PD MIC breakpoint was 16 mg/L in healthy volunteers and 8 to 12
mg/L (PTA = 87% for an MIC of 12 mg/L) in CF patients (Fig. 2C and D). For short-term (15-min) infusions of 2 g per 70 kg of WT q8h, the PK–PD MIC breakpoint was 6 mg/L in healthy volunteers and 3 mg/L in CF patients (Fig. 2A and B). This 2-fold difference in PK–PD MIC breakpoints for short-term intermittent treatment seems small. However, the dose in milligram-per-kilogram WT would have to be
136
J.B. Bulitta et al. / Diagnostic Microbiology and Infectious Disease 65 (2009) 130–141
Table 3 PK parameter estimates for the allometric body size model based on allometric scaling by LBM Parameter
Unit
Estimated for
CLTb CLR CLNR Vssb V1 V2 V3 CLicshallow CLicdeep TK0 (fixed) CVC SDC CVAU SDAU
L h−1 L h−1 L h−1 L L L L L h−1 L h−1 min % mg/L % mg
Medianf (90% confidence interval from nonparametric bootstrap)
Estimates from original dataset
CF patients
Healthy volunteers
6.33a,f 4.36 1.97 13.2 7.14 3.70 2.38 14.9 1.55 15 8.7 0.26 38 0.26
6.25f 4.54 1.71 12.7 6.87 3.56 2.29
Estimates for
Coefficient of variation (%)c
12 39 36d 18d 16d
16
CF patients
Healthy volunteers
6.34f (5.90–6.82) 4.38 (4.04–4.78) 1.97 (1.43–2.50) 12.8 (11.6–14.1) 6.88 (5.64–8.27) 3.42 (2.89–3.92) 2.45 (2.06–2.93) 14.3 (9.94–18.6) 1.58 (1.23–2.16) 15 8.7 (7.4–10) 0.23 (0.11–0.55) 36 (27–47) 0.26 (0.0026–1.18)
6.26f (5.92–6.63) 4.51 (4.19–4.86) 1.75 (1.46–2.05) 12.2 (11.5–13.0) 6.64 (5.56–7.59) 3.28 (2.77–3.76) 2.34 (1.97–2.81)
Coefficient of variation (%)c
11 (7.8–15) 36 (22–48) 34e (21–48) 21e (14–30) 16e (8.6–22)
16 (9.7–23)
CLT = total clearance; V1 = volume of distribution for the central compartment; V2 = volume of distribution for the shallow peripheral compartment; V3 = volume of distribution for the deep peripheral compartment; Vss = volume of distribution at steady state; CLicshallow = intercompartmental clearance between the central and the shallow peripheral compartment; CLicdeep = intercompartmental clearance between the central and the deep peripheral compartment; TK0 = duration of zero-order input (not estimated). CVC is the proportional and SDC is the additive residual error component for the plasma concentrations. CVAU is the proportional and SDAU is the additive residual error component for the amounts excreted in urine. a All clearance and volume terms are group estimates for subjects of standard body size (LBMSTD = 53 kg). b Derived from model estimates, not an estimated parameter. c Apparent coefficient of variation for the BSV. d The coefficients of correlation for pairs of random effects were rBSV(V1,V2) = −0.61, rBSV(V1,V3) = −0.22, and rBSV(V2,V3) = 0.06. e Median (90% confidence intervals) from bootstrap: rBSV(V1,V2) = −0.74 (−0.91 to −0.30), rBSV(V1,V3) = −0.11 (−0.52 to 0.31), and rBSV(V2,V3) = −0.048 (−0.55 to 0.38). f For the analysis of the original dataset, the estimates of structural model parameters are geometric means. For the bootstrap analysis, these estimates are the median of 1000 geometric means of the respective parameter.
doubled in CF patients, if CF patients should achieve the same PTAs as healthy volunteers in empiric therapy. We additionally explored dosing algorithms that aimed at achieving the same f T N MIC in CF patients of various body
sizes. These optimized dosing algorithms were based on an allometric scaling by LBM or WT and increased the PK–PD MIC breakpoints by about 5% to 15% compared with milligram-per-kilogram WT dosing for the 3 studied
Table 4 Ratio of group estimates (CF patients/healthy volunteers) for clearance and volume of distribution for different body size models Part A
Estimates from original dataset
Size model
FCYFCLR
FCYFCLNR
FCYFCLTa
FCYFVSS
1) 2) 3) 4) 5)
0.92 1.06 1.02 0.99 0.96
1.02 1.25 1.19 1.13 1.15
0.95 1.11 1.07 1.03 1.01
0.99 1.10 1.11 1.04 1.04
No size model WT linear scaling WT allometric LBM linear scaling LBM allometric
Part B
Median (90% confidence interval) from bootstrap replicates
Size model
FCYFCLR
FCYFCLNR
FCYFCLTa
FCYFVSS
1) 2) 3) 4) 5)
0.92 (0.84–1.00) 1.07 (0.95–1.22) 1.02 (0.92–1.15) 1.00 (0.89–1.13) 0.97 (0.87–1.09)
1.02 (0.75–1.39) 1.24 (0.88–1.66) 1.18 (0.86–1.59) 1.13 (0.82–1.50) 1.13 (0.79–1.49)
0.95 (0.87–1.04) 1.12 (1.02–1.23) 1.07 (0.98–1.17) 1.04 (0.94–1.14) 1.01 (0.93–1.10)
0.97 (0.85–1.09) 1.11 (1.00–1.26) 1.12 (1.00–1.27) 1.04 (0.95–1.14) 1.04 (0.95–1.15)
No size model WT linear scaling WT allometric LBM linear scaling LBM allometric
FCYFNNN = ratio of group estimates for parameter NNN after adjusting for body size and renal function (group estimate for CF patients divided by group estimate for healthy volunteers). a Calculated as FCYFCLT = (FCYFCLR · CLPOP,R + FCYFCLNR · CLPOP,NR)/(CLPOP,R + CLPOP,NR); the group estimates for renal and nonrenal clearance of healthy volunteers with standard body size are denoted as CLPOP,R and CLPOP,NR.
J.B. Bulitta et al. / Diagnostic Microbiology and Infectious Disease 65 (2009) 130–141 Table 5 Comparison of BSV (variances) between the different body size models Size model
Estimates from original dataset (%) CLR
2) WT linear scaling 3) WT allometric 4) LBM linear scaling 5) LBM allometric
a
100 62b,c 99c 70c
CLNR a
100 94 102 103
V1
V2 a
100 98 99 115
V3 a
100 99 85 73
100a 101 78 70
The table shows the variance for each body size model divided by the variance for linear scaling by WT for the respective PK parameter (Table 3 for parameter explanations). a The between-subject variances were scaled to the variance for linear scaling by WT. b Significantly smaller than 100% (P = 0.027 from 1000 bootstrap, 1sided test). All other 90% confidence intervals included 100%. c The lower this number, the more variability was explained by the respective body size model. These values mean that the BSV (variance) for renal clearance was reduced by 38% for allometric scaling by WT, by 1% for linear scaling by LBM, and by 30% for allometric scaling by LBM, all compared with linear scaling by WT.
carumonam dosage regimens (data not shown). Based on our population PK parameter estimates for healthy volunteers (Table 3), the expected PTA was 88% for continuous infusion, 86% for 4-h infusion q8h, and 65% for 15-min infusion for the MIC distribution in Japanese patients with urinary tract infections by P. aeruginosa at a daily dose of 6 g per 70 kg of WT. In MCS with larger variability in total clearance, the PK– PD MIC breakpoints for continuous and 4-h infusions q8h were approximately 1.5-fold lower for a %CV of 27% for total clearance and 2-fold lower for a %CV of 53% for total clearance compared with the breakpoints based on the final model (Table 2). For 15-min infusions q8h, PK–PD MIC breakpoints were 2-fold lower for a %CV of 27% for total clearance and 4-fold lower for a %CV of 53% for total clearance.
4. Discussion The improved life expectancy of CF patients has led to a significantly increased use of antibiotics in adult CF patients. However, only a few studies systematically compared the PK of antibiotics in CF patients and healthy volunteers. The ratio of mean clearance (in liters per hour per kilogram or liters per hour per 1.73 m2 body surface area) in CF patients divided by mean clearance in healthy volunteers ranges from 95% to 254% for β-lactams (Touw et al., 1998). This range is 64% to 144% for volume of distribution (in liters per kilogram or liters per 1.73 m2). A better understanding of these, in part, pronounced differences is important to adequately scale PK parameters by body size and optimize dosage regimens for CF patients of various body sizes. We studied various descriptors for body size (functional capacity) with linear or allometric scaling. A priori, we believed that it is important to account for differences in
137
body composition. Our study in adult subjects was rather small and had a relatively narrow range in body size. We incorporated body size and renal function as covariates based on prior knowledge, because covariate model building without prior knowledge is difficult for studies with small sample size (Ribbing and Jonsson, 2004). Carumonam was expected to primarily distribute into aqueous body compartments because β-lactams are hydrophilic molecules. Our CF patients and healthy volunteers had a similar LBM (Table 1). Thus, both groups were likely to have a comparable amount of aqueous body space and had a comparable volume of distribution (Table 2). Our PK parameter estimates were in good agreement with reports from other authors (Berube et al., 1988; Koeppe et al., 1987) for healthy volunteers. Unscaled renal clearance was 24% higher in CF patients (Table 2), which corresponded to the higher renal function in our CF patients compared with our healthy volunteers. This result is in agreement with studies in CF patients and healthy volunteers on other β-lactams (Bulitta et al., 2007; Hedman et al., 1988, 1990; Vinks et al., 2007; Wang et al., 1993). Renal function as a covariate on renal clearance explained the difference in average renal clearance of carumonam between CF patients and healthy volunteers and reduced the unexplained variance in renal clearance by 31% (from 14% to 12% CV). Therefore, renal function was a useful covariate in our study. However, our study was relatively small and is limited in representing the whole population of CF patients. In addition, Al-Aloul et al. (2007) pointed out limitations of the Cockcroft and Gault formula for predicting creatinine clearance in CF patients. Because our CF patients had a similar LBM compared with our healthy volunteers, the scale factor for clearance and volume of distribution was close to 1.0 (Table 4). CF patients had higher clearances and a larger volume of distribution for linear and allometric scaling by WT, probably because WT does not account for differences in body composition. Linear or allometric scaling by LBM explained the difference in average clearance and volume of distribution between both subject groups (Table 4). Allometric scaling by WT or LBM reduced the unexplained BSV by 30% to 38% for renal clearance, and allometric scaling by LBM reduced the unexplained BSV for volume of distribution of the peripheral compartments by about 30% (Table 5). Although renal function and LBM explained the difference in average PK parameters between both subject groups, CF patients still had a higher renal clearance than healthy volunteers. In agreement with the results for aztreonam from Vinks et al. (2007), CF patients achieved about 2 times lower PK–PD MIC breakpoints compared with healthy volunteers for intermittent short-term infusion with carumonam and dosing on a milligram-per-kilogram WT basis (Fig. 2). The lower PK–PD MIC breakpoints in CF patients were primarily a consequence 1) of the slightly higher renal clearance (and renal function) and 2) of the altered body composition of CF patients. It depends on the MIC distribution in the target patient population, whether or not a 2-fold difference in
138
J.B. Bulitta et al. / Diagnostic Microbiology and Infectious Disease 65 (2009) 130–141
Fig. 2. PTA for a daily carumonam dose of 6 g per 70 kg of WT at steady state for different targets of the PK–PD index fT N MIC.
PK–PD MIC breakpoints has a pronounced impact on the probability of successful treatment. Instead of increasing the dose, we studied prolonged and continuous infusion as alternative modes of administration to increase PK–PD MIC breakpoints. For the target f T N MIC ≥ 60%, we found a PK–PD MIC breakpoint of 8 to 12 mg/L (PTA = 87% at an MIC of 12 mg/L) in CF patients and of 16 mg/L in healthy volunteers for prolonged (4-h) infusions of 2 g per 70 kg of WT q8h (Fig. 2C and D). Continuous infusion at the same daily dose of 6 g per 70 kg of WT achieved a PK– PD MIC breakpoint of 16 mg/L in both subject groups. Dosing algorithms based on allometric scaling by LBM (Bulitta et al., 2006) yielded an additional improvement in the PK–PD MIC breakpoint by about 5% to 15% for the studied short-term, prolonged, and continuous infusion regimens.
Our estimates for the variability in renal and nonrenal clearance (Table 2) resulted in a %CV of 16% for total clearance. Because the variability of total clearance for βlactams in studies with CF patients (Touw et al., 1998) was larger than in our well-controlled study, we ran MCS with a %CV of 27% and 53% in total clearance. These are the median and 90% percentile of the variability in total clearance of 38 datasets on β-lactams (Touw et al., 1998). As expected, the PK–PD MIC breakpoints were lower (up to 4-fold for 15-min infusions q8h) compared with the MCS with the 16% CV in total clearance. To put these PTAs into clinical perspective, the expected PTA for treatment of infections caused by bacteria from a specific MIC distribution needs to be calculated (e.g., for the MIC distribution of a local hospital). Prolonged and
J.B. Bulitta et al. / Diagnostic Microbiology and Infectious Disease 65 (2009) 130–141
continuous infusions are expected to achieve notably higher PTAs than short-term infusions for some MICs. It is important to determine these MICs. Carumonam has good activity against Proteus mirabilis, K. pneumoniae, S. marcescens, Serratia liquefaciens, and E. coli with MIC90 values between ≤0.125 and 0.5 mg/L (Igari et al., 2003; Ikemoto et al., 1996a; Kumamoto et al., 2001, 2004, 2006). If the MIC90 is below 1 mg/L, short-term intermittent infusions at a daily dose of 6 g per 70 kg of WT were predicted to achieve a PTA expectation value N95%. However, other authors report a considerably lower susceptibility to carumonam. S. marcescens has been reported to have an MIC90 of 16 mg/L (Kumamoto et al., 2006), and P. aeruginosa had an MIC80 of 2 mg/L (Ikemoto et al., 1996b, 1997). More recent data on P. aeruginosa show an MIC90 of 32 mg/L, with 61% of the isolates having an MIC of 2 to 16 mg/L (Kumamoto et al., 2006). It is the infections caused by these pathogens for which prolonged and continuous infusion will have considerably higher PTA expectation values compared with short-term infusion. This applies especially to CF patients because they had a breakpoint of 3 mg/L for short-term infusions of 2 g per 70 kg of WT q8h and a breakpoint of 8 to 12 mg/L for prolonged infusions at the same dose. Two limitations of our MCS for CF patients are that we did not measure concentrations of carumonam in the lungs of CF patients and that we are not aware of an established PK– PD target for successful therapy in CF patients. Especially for CF patients with chronic infections by P. aeruginosa, it seems likely that substantially higher PK–PD targets and possibly also combination therapy are required for successful treatment compared with early P. aeruginosa infections (Gibson et al., 2003). In conclusion, we found a 24% higher unscaled renal clearance in CF patients compared with healthy volunteers that could be explained by an enhanced glomerular filtration rate in CF patients. The difference in average PK parameters between CF patients and healthy volunteers could be explained by body size, body composition, and renal function. These differences were better explained by LBM than by WT. Allometric scaling by LBM reduced the unexplained BSV in renal clearance and volume of distribution of both peripheral compartments by about 30% relative to linear scaling by WT. Future clinical studies are warranted to explore the expected higher probability of successful clinical outcome for allometric dose selection in CF patients based on LBM. Prolonged and continuous infusion achieved higher PK–PD MIC breakpoints compared with short-term infusions q8h at the same daily dose. At a daily dose of 6 g per 70 kg of WT, CF patients achieved a PK–PD MIC breakpoint (target: f T N MIC ≥ 60%) of 3 mg/L for 15-min infusions q8h, of 8 to 12 mg/L (87% PTA for MIC = 12 mg/L) for 4-h infusions q8h, and of 16 mg/L for continuous infusion. Therefore, prolonged infusion was predicted to be especially superior to short-term infusion, if the MIC50 and MIC90 fall between 2 and 8 mg/L.
139
Acknowledgment The authors thank Dr Nick Holford and Dr Thomas Lodise for very fruitful discussions and comments on this project and manuscript. References Al-Aloul M, Jackson M, Bell G, Ledson M, Walshaw M (2007) Comparison of methods of assessment of renal function in cystic fibrosis (CF) patients. J Cyst Fibros 6:41–47. Anderson BJ, Holford NH (2008) Mechanism-based concepts of size and maturity in pharmacokinetics. Annu Rev Pharmacol Toxicol 48: 303–332. Beal SL, Boeckmann AJ, Sheiner LB, NONMEM Project Group (1999) NONMEM users guides. San Francisco: University of California at San Francisco. Berube D, Vallee F, Panneton AC, Bergeron MG, LeBel M (1988) Pharmacokinetics of carumonam after single and multiple 1- and 2-g dosage regimens. Antimicrob Agents Chemother 32:354–357. Bulitta JB, Duffull SB, Landersdorfer CB, Drusano GL, Kinzig-Schippers M, Holzgrabe U, Stephan U, Sörgel F (2006) Pharmacodynamic comparison of cystic fibrosis patients and healthy volunteers by population pharmacokinetics (PK) and Monte Carlo simulation. Poster W4056. AAPS Annual Meeting, San Antonio, TX, USA. Bulitta JB, Duffull SB, Kinzig-Schippers M, Holzgrabe U, Stephan U, Drusano GL, Sorgel F (2007) Systematic comparison of the population pharmacokinetics and pharmacodynamics of piperacillin in cystic fibrosis patients and healthy volunteers. Antimicrob Agents Chemother 51:2497–2507. Cheymol G (2000) Effects of obesity on pharmacokinetics implications for drug therapy. Clin Pharmacokinet 39:215–231. Cockcroft DW, Gault MH (1976) Prediction of creatinine clearance from serum creatinine. Nephron 16:31–41. Craig WA (1998) Pharmacokinetic/pharmacodynamic parameters: rationale for antibacterial dosing of mice and men. Clin Infect Dis 26:1–12. Craig WA (2003) Basic pharmacodynamics of antibacterials with clinical applications to the use of β-lactams, glycopepq8hes, and linezolid. Infect Dis Clin North Am 17:479–501. Drusano GL (2004) Antimicrobial pharmacodynamics: critical interactions of “bug and drug”. Nat Rev Microbiol 2:289–300. Efron B, Tibshirani RJ (1986) Bootstrap methods for standard errors, confidence intervals and other measures of statistical accuracy. Stat Sci 1:54–77. Gibson RL, Burns JL, Ramsey BW (2003) Pathophysiology and management of pulmonary infections in cystic fibrosis. Am J Respir Crit Care Med 168:918–951. Gottschalk B, Sorgel F, Wiesemann HG, Stephan U, Rey M, Bowing HB, Stehr C, Dominick HC, Kiosz D, Geldmacher-von Mallinckrodt M (1988) Antibiotic pharmacokinetics in patients with mucoviscidosis. Monatsschr Kinderheilkd 136:475–478. Green B, Duffull SB (2004) What is the best size descriptor to use for pharmacokinetic studies in the obese? Br J Clin Pharmacol 58:119–133. Hedman A, Adan-Abdi Y, Alvan G, Strandvik B, Arvidsson A (1988) Influence of the glomerular filtration rate on renal clearance of ceftazidime in cystic fibrosis. Clin Pharmacokinet 15:57–65. Hedman A, Alvan G, Strandvik B, Arvidsson A (1990) Increased renal clearance of cefsulodin due to higher glomerular filtration rate in cystic fibrosis. Clin Pharmacokinet 18:168–175. Holford NH (1996) A size standard for pharmacokinetics. Clin Pharmacokinet 30:329–332. Igari J, Oguri T, Hiramatsu N, Akiyama K, Koyama T (2003) Postmarketing surveillance of antibacterial activities of cefozopran against various clinical isolates—II. Gram-negative bacteria. Jpn J Antibiot 56: 458–496.
140
J.B. Bulitta et al. / Diagnostic Microbiology and Infectious Disease 65 (2009) 130–141
Ikemoto H, Watanabe K, Mori T, Igari J, Oguri T, Kobayashi K, Satou K, Matsumiya H, Saito A, Terai T, Tanno Y, Nishioka K, Arakawa M, Wada K, Okada M, Ozaki K, Aoki N, Kitamura N, Sekine O, Suzuki Y, Matsuda M, Tanimoto H, Nakata K, Nakamori Y, Kusano N, et al. (1996a) Susceptibilities of bacteria isolated from patients with respiratory infectious diseases to antibiotics (1994). Jpn J Antibiot 49:419–455. Ikemoto H, Watanabe K, Mori T, Igari J, Oguri T, Kobayashi K, Satou K, Matsumiya H, Saito A, Terai T, Tanno Y, Nishioka K, Arakawa M, Wada K, Okada M, Ozaki K, Aoki N, Kitamura N, Sekine O, Suzuki Y, Tanimoto H, Nakata K, Nakamori Y, Nakatani T, Kusano N, et al. (1996b) Susceptibilities of bacteria isolated from patients with respiratory infectious diseases to antibiotics (1992). Jpn J Antibiot 49:34–70. Ikemoto H, Watanabe K, Mori T, Igari J, Oguri T, Kawaguchi H, Shimizu Y, Matsumiya H, Saito A, Terai T, Inoue H, Nakadate T, Ito C, Yosida T, Tanno Y, Ohno I, Nishioka K, Arakawa M, Igarashi K, Wada K, Okada M, Ozaki K, Aoki N, Kitamura N, Touyama M, et al. (1997) Susceptibilities of bacteria isolated from patients with respiratory infectious diseases to antibiotics (1995). Jpn J Antibiot 50:421–459. Ikemoto H, Watanabe K, Mori T, Igari J, Oguri T, Shimizu Y, Terai T, Inoue H, Nakadate T, Ito C, Yoshida T, Ohno I, Tanno Y, Arakawa M, Igarashi K, Okada M, Ozaki K, Aoki N, Kitamura N, Sekine O, Suzuki Y, Nakata K, Nakatani T, Inagawa H, Kusano N (1998) Susceptibilities of bacteria isolated from patients with lower respiratory infectious diseases to antibiotics (1996). Jpn J Antibiot 51:437–474. Imada A, Kondo M, Okonogi K, Yukishige K, Kuno M (1985) In vitro and in vivo antibacterial activities of carumonam (AMA-1080), a new N-sulfonated monocyclic beta-lactam antibiotic. Antimicrob Agents Chemother 27:821–827. James W (1976) Research on Obesity. London: Her Majesty's Stationery Office. Jones RN, Barry AL, Thornsberry C, Fuchs PC, Packer RR (1986) The antimicrobial activity, beta-lactamase stability, and disk diffusion susceptibility testing of carumonam (RO 17-2301, AMA-1080), a new monobactam. Am J Clin Pathol 86:608–618. Koeppe P, Hoffler D, Strobel K (1987) Pharmacokinetics and dose recommendations of carumonam in renal failure. Arzneimittelforschung 37:65–69. Kumamoto Y, Tsukamoto T, Watanabe K, Kobayashi Y, Hirose T, Matsukawa M, Uchida H, Takahashi S, Kunishima Y, Matsuda S, Sato S, Ogihara M, Ishibashi K, Shigeta S, Fujime M, Fujita K, Yoshida H, Imafuku Y, Igari J, Murai M, Oguri T, Yamaguchi K, Matsumoto T, Kashitani F, Furuhama T, Kumon H, Monden K, Ooe H, Nishikawa M, Kohno S, Miyazaki Y, Izumikawa K, Yamaguchi T, Mochida C, Oka T, Kitamura M, Takano Y, Matsuoka Y (2001) Comparative studies on activities of antimicrobial agents against causative organisms isolated from patients with urinary tract infections (1999). I. Susceptibility distribution. Jpn J Antibiot 54:185–216. Kumamoto Y, Tsukamoto T, Watanabe K, Kobayashi Y, Matsukawa M, Uchida H, Kunishima Y, Hirose T, Matsuda S, Sato S, Shigeta S, Fujime M, Fujita K, Yamaguti O, Ogihara M, Ishibashi K, Igari J, Takahashi K, Oguri T, Yoshida H, Imafuku Y, Yamaguchi K, Furuya N, Kashitani F, Murai M, Ooe H, Nishikawa M, Kumon H, Monden K, Oka T, Kitamura M, Kohno S, Tomono K, Miyazaki Y, Matsuoka Y, Fukuhara Y, Hirakata Y, Aoki S (2003) Comparative studies on activities of antimicrobial agents against causative organisms isolated from patients with urinary tract infections (2001). I. Susceptibility distribution. Jpn J Antibiot 56: 396–423. Kumamoto Y, Tsukamoto T, Matsukawa M, Kunishima Y, Hirose T, Yamaguti O, Ishibashi K, Shigeta S, Suzutani T, Yoshida H, Imafuku Y, Murai M, Watanabe K, Kobayashi Y, Uchida H, Matsuda S, Sato S, Fujime M, Fujita K, Igari J, Oguri T, Yamaguchi K, Furuya N, Deguchi T, Ishihara S, Ooe H, Nishikawa M, Oka T, Kitamura M, Fukuhara Y, Kamidono S, Arakawa S, Kumon H, Monden K, Matsumoto T, Takahashi K, Naito S, Egashira T, Kohno S, Miyazaki Y, Hirakata Y, Aoki S (2004) Comparative studies on activities of antimicrobial agents against causative organisms isolated from patients with urinary tract infections (2002). I. Susceptibility distribution. Jpn J Antibiot 57:246–274.
Kumamoto Y, Tsukamoto T, Matsukawa M, Kunishima Y, Hirose T, Shigeta S, Yamaguti O, Ishibashi K, Suzutani T, Yoshida H, Imafuku Y, Murai M, Watanabe K, Kobayashi Y, Uchida H, Matsuda S, Sato S, Fujime M, Fujita K, Igari J, Oguri T, Yamaguchi K, Furuya N, Deguchi T, Ishihara S, Ooe H, Oka T, Kitamura M, Fukuhara Y, Kamidono S, Arakawa S, Kumon H, Monden K, Matsumoto T, Muratani T, Naito S, Egashira T, Konishi T, Kohno S, Hirakata Y, Kondo A, Matsuda J, Nakano M (2006) Comparative studies on activities of antimicrobial agents against causative organisms isolated from patients with urinary tract infections (2004). I. Susceptibility distribution. Jpn J Antibiot 59: 177–200. McNulty CA, Garden GM, Ashby J, Wise R (1985) Pharmacokinetics and tissue penetration of carumonam, a new synthetic monobactam. Antimicrob Agents Chemother 28:425–427. Miller ME, Kornhauser DM (1994) Bromide pharmacokinetics in cystic fibrosis. Arch Pediatr Adolesc Med 148:266–271. Morgan DJ, Bray KM (1994) Lean body mass as a predictor of drug dosage. Implications for drug therapy. Clin Pharmacokinet 26:292–307. Mosteller RD (1987) Simplified calculation of body-surface area. N Engl J Med 317:1098. Paap CM, Nahata MC (1990) Clinical pharmacokinetics of antibacterial drugs in neonates. Clin Pharmacokinet 19:280–318. Parke J, Holford NH, Charles BG (1999) A procedure for generating bootstrap samples for the validation of nonlinear mixed-effects population models. Comput Methods Programs Biomed 59:19–29. Prandota J (1987) Drug disposition in cystic fibrosis: progress in understanding pathophysiology and pharmacokinetics. Pediatr Infect Dis J 6:1111–1126. Prandota J (1988) Clinical pharmacology of antibiotics and other drugs in cystic fibrosis. Drugs 35:542–578. Raimondi A, Mattina R, Cocuzza CE (1988) Carumonam's in-vitro activity against gram-negative bacteria and its stability to their beta-lactamases. Chemioterapia 7:3–9. Rey E, Treluyer JM, Pons G (1998) Drug disposition in cystic fibrosis. Clin Pharmacokinet 35:313–329. Ribbing J, Jonsson EN (2004) Power, selection bias and predictive performance of the Population Pharmacokinetic Covariate Model. J Pharmacokinet Pharmacodyn 31:109–134. Rochat T, Slosman DO, Pichard C, Belli DC (1994) Body composition analysis by dual-energy x-ray absorptiometry in adults with cystic fibrosis. Chest 106:800–805. Shinagawa N, Yura J, Manabe T, Mashita K, Ishikawa S, Mizuno A, Hirata K, Denno R, Mukaiya M, Ishibiki K, Ushijima Y, Aikawa N, Takuma K, Iwai S, Kunimatsu M, Ohtsuka K, Kinoshita H, Morimoto K, Fujimoto M, Tanimura H, Ohnishi H, Umemoto Y, Sakaguchi S, Dounishi H, Oda S, et al. (1996) Isolation rate of Pseudomonas aeruginosa from surgical infections and their susceptibilities. Jpn J Antibiot 49:544–554. Slosman DO, Casez JP, Pichard C, Rochat T, Fery F, Rizzoli R, Bonjour JP, Morabia A, Donath A (1992) Assessment of whole-body composition with dual-energy x-ray absorptiometry. Radiology 185:593–598. Sorgel F, Stephan U, Wiesemann HG, Gottschalk B, Stehr C, Rey M, Bowing HB, Dominick HC, Geldmacher von Mallinckrodt M (1987) High dose treatment with antibiotics in cystic fibrosis—a reappraisal with special reference to the pharmacokinetics of β-lactams and new fluoroquinolones in adult CF-patients. Infection 15:385–396. Sweetman SC (2009) Martindale: the complete drug reference [online]. London: Pharmaceutical Press. pp. 36. http://www.medicinescomplete. com/. Touw DJ (1998) Clinical pharmacokinetics of antimicrobial drugs in cystic fibrosis. Pharm World Sci 20:149–160. Touw DJ, Vinks AA, Heijerman HG, Hermans J, Bakker W (1994) Suggestions for the optimization of the initial tobramycin dose in adolescent and adult patients with cystic fibrosis. Ther Drug Monit 16: 125–131. Touw DJ, Vinks AA, Heijerman HG, Bakker W (1996) Prospective evaluation of a dose prediction algorithm for intravenous tobramycin in
J.B. Bulitta et al. / Diagnostic Microbiology and Infectious Disease 65 (2009) 130–141 adolescent and adult patients with cystic fibrosis. Ther Drug Monit 18: 118–123. Touw DJ, Vinks AA, Mouton JW, Horrevorts AM (1998) Pharmacokinetic optimisation of antibacterial treatment in patients with cystic fibrosis. Current practice and suggestions for future directions. Clin Pharmacokinet 35:437–459. Vinks AA, van Rossem RN, Mathot RA, Heijerman HG, Mouton JW (2007) Pharmacokinetics of aztreonam in healthy subjects and patients with cystic fibrosis and evaluation of dose–exposure relationships using Monte Carlo simulation. Antimicrob Agents Chemother 51: 3049–3055. Wang JP, Unadkat JD, al-Habet SM, O'Sullivan TA, Williams-Warren J, Smith AL, Ramsey B (1993) Disposition of drugs in cystic fibrosis. IV.
141
Mechanisms for enhanced renal clearance of ticarcillin. Clin Pharmacol Ther 54:293–302. West GB, Brown JH, Enquist BJ (1997) A general model for the origin of allometric scaling laws in biology. Science 276:122–126. West GB, Brown JH, Enquist BJ (1999) The fourth dimension of life: fractal geometry and allometric scaling of organisms. Science 284:1677–1679. Yoshida K, Mitani M, Naeshiro I, Torii H, Tanayama S (1986) Disposition of carumonam (AMA-1080/Ro 17-2301), a new N-sulfonated monocyclic β-lactam, in rats and dogs. Antimicrob Agents Chemother 29: 1017–1024. Zhao WH, Hu ZQ, Shimura T (2008) Potency of IMP-10 metallo-betalactamase in hydrolysing various antipseudomonal beta-lactams. J Med Microbiol 57:974–979.
Diagnostic Microbiology and Infectious Disease 65 (2009) 130 – 141 www.elsevier.com/locate/diagmicrobio
Comparison of the pharmacokinetics and pharmacodynamic profile of carumonam in cystic fibrosis patients and healthy volunteers☆ Jürgen B. Bulittaa,1 , Stephen B. Duffullb , Cornelia B. Landersdorfera,1 , Martina Kinziga , Ulrike Holzgrabec , Ulrich Stephana,d , George L. Drusanoe , Fritz Sörgela,d,⁎ a
IBMP—Institute for Biomedical and Pharmaceutical Research, D-90562 Nürnberg-Heroldsberg, Germany b School of Pharmacy, University of Otago, 9054 Dunedin, New Zealand c Institute of Pharmacy and Food Chemistry, University of Würzburg, D-97074 Würzburg, Germany d Department of Pharmacology, University of Duisburg–Essen, D-45122 Essen, Germany e Ordway Research Institute, Albany, NY 12208, USA Received 31 March 2008; accepted 18 June 2009
Abstract Our objectives were to compare the pharmacokinetics (PK) of carumonam, a monobactam, between cystic fibrosis (CF) patients and healthy volunteers and assess its pharmacodynamic profile. We studied 10 adult CF patients and 18 healthy volunteers of similar body size (dose: 2.166 g of carumonam as 15-min intravenous infusion). High performance liquid chromatography with ultraviolet detection (HPLCUV) was used for drug analysis and NONMEM® (ICON, Ellicot City, MD) for population PK and Monte Carlo simulation with targets between ≥20% and 100% free time above MIC (f T N MIC). Unscaled renal clearance was 24% higher in CF patients. Lean body mass and creatinine clearance explained the difference in average clearance and volume of distribution between both subject groups. For a daily dose of 6 g per 70 kg of total body weight, 15-min infusions q8h achieved robust (N90%) probabilities of target attainment (PTAs) (target, 60% f T N MIC) for MICs ≤3 mg/L in CF patients and ≤6 mg/L in healthy volunteers. At the same dose, 4-h infusions q8h achieved robust PTAs up to markedly higher MICs ≤8 to 12 mg/L in CF patients and ≤16 mg/L in healthy volunteers. © 2009 Elsevier Inc. All rights reserved. Keywords: Cystic fibrosis/healthy volunteers; Population pharmacokinetics/pharmacodynamics of carumonam; Monte Carlo simulation; Allometric scaling by lean body mass; PK–PD MIC breakpoints
1. Introduction In Caucasians, cystic fibrosis (CF) is the most common inherited disease. Abnormal chloride transport across the apical membrane of epithelial cells of CF patients causes a high degree of morbidity and mortality. The most serious ☆
This work was presented in part as part of a pharmacokinetic metaanalysis at the 45th Interscience Conference on Antimicrobial Agents and Chemotherapy, 2005 (poster abstract A-12), as part of a pharmacodynamic meta-analysis at the AAPS Annual Meeting, 2006 (poster W4056), and as a chapter in J. Bulitta's PhD thesis. ⁎ Corresponding author. IBMP—Institute for Biomedical and Pharmaceutical Research, D-90562 Nürnberg-Heroldsberg, Germany. Tel.: +49911-518-290; fax: +49-911-518-2920. E-mail address: [email protected] (F. Sörgel). 1 Present address: Department of Pharmaceutical Sciences, State University of New York at Buffalo, NY 14260, USA. 0732-8893/$ – see front matter © 2009 Elsevier Inc. All rights reserved. doi:10.1016/j.diagmicrobio.2009.06.018
hazard is inspissated secretions in the airways. Those secretions host many bacteria that can cause respiratory tract infections. The high morbidity and mortality involved in this pulmonary disease underline the need for optimal antiinfective treatment of CF patients. After initial infection by Haemophilus influenzae or Staphylococcus aureus, many CF patients experience infections by Pseudomonas aeruginosa, methicillin-resistant S. aureus, or Burkholderia cepacia. Therefore, it is important to predict and optimize the probability of successful antibiotic treatment against those pathogens as a function of their MIC distribution in CF patients. Carumonam is an N-sulfonated monobactam antibiotic that is relatively stable against β-lactamases (Imada et al., 1985; Jones et al., 1986; Raimondi et al., 1988; Zhao et al., 2008) and is marketed as Amasulin® by Takeda in Japan. Carumonam is active against H. influenzae (Ikemoto et al.,
J.B. Bulitta et al. / Diagnostic Microbiology and Infectious Disease 65 (2009) 130–141
1998), Klebsiella pneumoniae (Kumamoto et al., 2006), P. aeruginosa (Ikemoto et al., 1996b, 1997; Shinagawa et al., 1996), Escherichia coli (Kumamoto et al., 2006), and Serratia marcescens (Igari et al., 2003; Kumamoto et al., 2003) and might be a treatment option for respiratory or urinary tract infections caused by those pathogens (Yoshida et al., 1986). Typical daily doses range from 1 to 2 g of carumonam split into 2 doses (Sweetman, 2009). Doses of 2 g q8h were well tolerated in healthy volunteers (Berube et al., 1988). We are not aware of any study on the pharmacokinetics (PK) of carumonam in CF patients. For many antibiotics, there are too few controlled clinical trials that systematically compare the PK of antibiotics in CF patients and healthy volunteers (Gottschalk et al., 1988; Prandota, 1987, 1988; Rey et al., 1998; Sorgel et al., 1987; Touw, 1998; Touw et al., 1998). This lack of data may hamper optimal dose selection in CF patients. Many PK studies in CF patients lack a control group of healthy volunteers, which makes a PK comparison of CF patients and healthy volunteers difficult. One difference between CF patients and healthy volunteers is that CF patients are, on average, leaner than healthy volunteers. Body size and body composition have been shown to influence PK (Cheymol, 2000; Green and Duffull, 2004; Morgan and Bray, 1994; Paap and Nahata, 1990; Touw et al., 1998). This requires dose adjustment in some patient groups. Because CF patients are often lean, their lean body mass (LBM) makes up a larger proportion of their total body weight (WT) compared with healthy volunteers (Rochat et al., 1994; Slosman et al., 1992). From a PK point of view, the ideal size descriptor should be able to describe body size and body composition for a large variety of patient groups, for example, for obese, undernourished, and “normal” patients. The ideal size descriptor should also reduce the unexplained (random) between-subject variability (BSV) in PK parameters because this would allow one to achieve target concentrations more precisely in empiric therapy. Population PK can directly use body size, body composition, and renal function to explain the observed variability in PK parameters between patients (e.g., for clearance and volume of distribution). The ability of various body size models to describe the difference in average clearance and volume of distribution and to reduce the unexplained BSV can be directly compared via population PK. There are reports that clearance and volume of distribution can be better described by LBM than by WT in CF patients (Miller and Kornhauser, 1994; Prandota, 1987; Rey et al., 1998; Touw et al., 1994, 1996, 1998). Touw et al. (1994, 1996) proposed to use LBM instead of WT for initial dose selection of tobramycin in CF patients. Although LBM has been proposed to be a superior predictor for dose selection compared with WT or body surface area (Morgan and Bray, 1994), the potential merits of dose selection based on LBM in CF patients still need to be shown (Morgan and Bray, 1994; Rey et al., 1998). Our primary objective was to compare the population PK of carumonam between CF patients and healthy volunteers
131
and to study whether LBM is a more appropriate descriptor of body size and functional capacity than WT. As secondary objective, we assessed the probability of target attainment (PTA) for CF patients and healthy volunteers for various dosage regimens based on PK–pharmacodynamic (PD) targets via Monte Carlo simulation (MCS). We used the time during which the non–protein-bound plasma concentration exceeds the minimal inhibitory concentration (f T N MIC) in combination with PK–PD targets between 20% and 100% f T N MIC. 2. Materials and methods 2.1. Subjects A total of 28 Caucasian volunteers (10 CF patients and 18 healthy volunteers) of both sexes participated in the study after they had given their written informed consent. The subjects' health status was assessed by physical examination, electrocardiography, and laboratory tests including urinalysis and screening for drugs of abuse. Consumption of alcohol in any form or of other medication (including antibiotics) was forbidden from 24 h before carumonam dose until the last sample. After fasting overnight, the subjects received a standardized breakfast at 1 h postdose and a standardized lunch at 4 h postdose. Fluid intake was standardized to 125 mL of mineral water per hour. All volunteers were closely observed by physicians for the occurrence of adverse events during the period of drug administration. The study protocol had been approved by the local ethics committee, and the study was performed according to the revised version of the Declaration of Helsinki. 2.2. Study design and drug administration The study was a single-dose, single-center, open, parallel group study. Each subject received a dose of 2.166 g of carumonam as a 15-min intravenous infusion. All infusions were administered with a BRAUN-Perfusor® (Braun, Melsungen, Germany). The instruments were checked on a daily basis by weighing defined volumes delivered by the Perfusor. 2.3. Blood and urine sampling All blood samples were drawn in vials containing citric acid from a forearm vein via an intravenous catheter contralateral to the one used for drug administration. Blood samples were drawn immediately before start of the infusion (0 min) and at 5, 10, and 15 min poststart of infusion as well as at 5, 15, 20, 30, 45, 60, 75, and 90 min and 2, 3, 4, 5, 6, 8, 10, 12, and 24 h after the end of infusion. After centrifugation, all plasma samples were immediately frozen and stored at −70 °C until analysis. Urine was collected from the start of the infusion until 1 h after the end of infusion as well as at 1 to 2, 2 to 3, 3 to 4, 4 to 6, 6 to 8, 8 to 10, 10 to 12, and 12 to 24 h after end of infusion. The urine samples were
132
J.B. Bulitta et al. / Diagnostic Microbiology and Infectious Disease 65 (2009) 130–141
stored at 4 °C during the collection period. The amount and pH of the urine were measured, and aliquots were frozen and stored at −70 °C until analysis.
The same allometric body size model was applied for LBM with a standard LBM (LBMSTD) of 53 kg.
2.4. Drug analysis in plasma and urine
FSize;V;i =
The concentration of carumonam in plasma and urine was determined by high-performance liquid chromatography. Plasma samples (200 μL) were deproteinized by addition of 400 μL of acetonitrile. This mixture was centrifuged at 10,000 rpm for 5 min, and 1000 μL of dichloromethane were added to the supernatant. Ten microliters of the aqueous phase were injected onto the system. We used a Hypersil ODS 5 (Thermo Fisher Scientific, Dreieich, Germany), 125 × 4 mm column and a 1% acetic acid (glacial)/2% methanol, 0.3% (NH4)2SO4 eluent at pH 2.9. Carumonam was detected at a wavelength of 293 nm. The limit of quantification was 0.2 mg/L for plasma and 7 mg/L for urine. The coefficients of variation for interday precision were between 4.1% and 5.5% for plasma and between 2.6% and 6.0% for urine.
LBMi LBMSTD LBMi 0:75 FSize; CL;i = LBMSTD
ð3Þ ð4Þ
All exponents were set to 1.0 for linear scaling by WT or linear scaling by LBM. 2.5.3. Renal function The glomerular filtration rate of the ith subject (GFRi) was estimated by the Cockcroft and Gault (1976) formula for a nominal 70-kg subject: " agei GFRi = 140 years # mg=dL mL ð5Þ Fsex serum creatinine concentrationi min
2.5. Population PK analysis 2.5.1. Structural model One-, 2-, and 3-compartment disposition models with a time-delimited zero-order input into the central compartment were tested. Competing models were discriminated by their predictive performance assessed via visual predictive checks, NONMEM's objective function (proportional to −2 times the log-likelihood), and residual plots. 2.5.2. Size We studied the following body size models: 1) no body size model, 2) linear scaling by WT, 3) allometric scaling by WT (Anderson and Holford, 2008; Holford, 1996; West et al., 1997, 1999), 4) linear scaling by LBM (Cheymol, 2000; James, 1976), and 5) allometric scaling by LBM. The ability of the body size models to describe the differences in the central tendency of PK parameters was compared between CF patients and healthy volunteers. Furthermore, we compared the ability of the body size models to reduce the unexplained BSV. Our allometric model assumes that volume of distribution scales linearly (allometric exponent 1.0) with body size (i.e., WT or LBM) and that clearance scales slightly less than linearly with body size (allometric exponent, 0.75). The allometric exponent was fixed to 1.0 for all volume terms and fixed to 0.75 for all clearance terms. The FSize,V,i and FSize,CL,i are the fractional changes in volume of distribution and clearance for the ith subject (with WTi) standardized to a weight WTSTD of 70 kg. FSize;V;i =
WTi WTSTD
WTi FSize; CL;i = WTSTD
ð1Þ 0:75
The sex-specific factor (Fsex) is 0.85 for females and 1.0 for males. We scaled the estimated glomerular filtration rate by a standard glomerular filtration rate (GFRSTD) of 120 mL/min per 70 kg. Renal function of the ith subject (RFi) can then be predicted as follows: RFi =
GFRi GFRSTD
ð6Þ
2.5.4. Clearance We assumed a nonrenal and a renal component of clearance (CLNR and CLR, respectively). Renal clearance of carumonam was estimated based on the amount of carumonam excreted unchanged into urine, and renal function, (RF) was used as a covariate on renal clearance. Renal clearance was assumed to be linearly related to RF. 2.5.5. BSV model We estimated the BSV for renal clearance, nonrenal clearance, volume of distribution, and duration of zero-order input by assuming a log-normal distribution for the PK parameters. The log-scale differences (ηBSV) of the individual PK parameters from their population mean are assumed to be normally distributed random variables with mean zero and standard deviation BSV. The BSV was estimated as variance, but we report the square root of the estimate. We expressed these values in the text as a percentage, because this quantity is an approximation to the apparent coefficient of variation of a normal distribution on log scale. The individual PK parameters were calculated as follows (parameters explained below): CLR;i = CLPOP;R FSize;CL;i FCYFCLR RFi expðgBSVCLRi Þ
ð2Þ
ð7Þ
J.B. Bulitta et al. / Diagnostic Microbiology and Infectious Disease 65 (2009) 130–141
CLNR;i = CLPOP;NR FSize;CL;i FCYFCLNR expðgBSVCLNRi Þ ð8Þ CLi = CLR;i + CLNR;i
ð9Þ
The CLi, CLRi, and CLNRi are the individual estimates for total, renal, and nonrenal clearance for the ith subject. The ηBSVCLRi and ηBSVCLNRi are ηBSV of CLR and CLNR for the ith subject, and CLPOP,R and CLPOP,NR are the group estimates for the renal and nonrenal clearance of a healthy volunteer with standard body size (i.e., WTSTD = 70 kg or LBMSTD = 53 kg) and with an RF of 1. The FCYFCLR and FCYFCLNR represent disease-specific scale factors for CF patients that characterize the difference in renal and nonrenal clearance between CF patients and healthy volunteers. A disease-specific scale factor for volume of distribution at steady state (FCYFVSS) was also estimated. 2.5.6. Nonparametric bootstrap Confidence intervals of population PK parameter estimates were determined by nonparametric bootstrap resampling techniques with 1000 replicates for each body size model (Efron and Tibshirani, 1986; Parke et al., 1999). Data from 10 randomly selected CF patients and 18 randomly selected healthy volunteers from the original raw dataset (subjects could be drawn multiple times with replacement) were included in each replicate. The median and nonparametric 90% confidence intervals (5–95% percentile) were calculated from the 1000 estimated PK parameters of our final population PK model. 2.5.7. Observation model and computation The residual unidentified variability was described by a combined additive and proportional error model for plasma concentrations and amounts excreted in urine. We used the first-order conditional estimation method with the interaction estimation option in NONMEM version V release 1.1 (NONMEM Project Group, University of California, San Francisco, CA) (Beal et al., 1999) for all population PK modeling. 2.5.8. Noncompartmental analysis The vast majority of PK studies in CF patients was analyzed by noncompartmental methods (Rey et al., 1998; Touw et al., 1998). For easier comparison of our study with studies in literature, we used WinNonlin™ Professional (version 4.0.1; Pharsight, Mountain View, CA) to calculate noncompartmental PK parameters and descriptive statistics. 2.5.9. Monte Carlo simulation Because carumonam is a β-lactam, we assumed that f T N MIC is the most appropriate PK–PD index. We calculated the PTA for f T N MIC targets of at least 20%, 40%, 60%, 80%, or 100% within the MIC range from 0.25 to 64 mg/L. A protein binding of 23% was used for carumonam, which was shown to be independent of plasma concentration between 25 and 400 mg/L (McNulty et al., 1985).
133
We compared the PTA versus MIC profiles for a daily carumonam dose of 6 g per 70 kg of WT given 1) as continuous infusion, 2) split into three 15-min infusions q8h, or 3) split into three 4-h infusions q8h. Concentration time profiles of 10,000 CF patients and 18,000 healthy volunteers with the same demographic data as the subjects in this study were simulated at steady state in absence of residual error in NONMEM. We used the population mean parameter estimates and parameter variability model of our final population PK model with a block diagonal variance– covariance matrix for MCS; dose, timing of infusion, and protein binding were assumed to have no variability. The f T N MIC values and the PTA were calculated by linear interpolation between simulated data points (19 observations per 8 h) by validated Perl scripts. We derived the PTA for CF patients and healthy volunteers by calculating the fraction of subjects who attained the PK–PD target at each MIC for each target. The PK–PD MIC breakpoint was defined as the highest MIC for which the PTA was at least 90%. The PK–PD MIC breakpoints for the target f T N MIC ≥ 60% are reported in the text because the target for near-maximal bactericidal activity for other β-lactams fall between 40% and 70% f T N MIC (Craig, 1998, 2003; Drusano, 2004). The PTA versus MIC profiles are presented for targets of 20%, 40%, 60%, 80%, and 100% f T N MIC in graphical form to cover a wide range of PK–PD targets. We ran additional simulations based on the individual PK parameter estimates and covariates (WT and LBM) from the 10 CF patients in our study to compare dosing algorithms based on the body size models identified in this analysis as providing the best biologic description of the data (Bulitta et al., 2006). The expected PTA was calculated for Japanese patients with urinary tract infections caused by P. aeruginosa based on their MIC distribution from 2004 (Kumamoto et al., 2006). These MICs ranged from 0.25 to 128 mg/L (MIC50, 4 mg/L; MIC90, 32 mg/L). For this MCS, a daily carumonam dose of 6 g per 70 kg of WT and our final population PK parameter estimates for healthy volunteers were used. To reflect the potentially larger variability in a broader (and potentially more sick) CF patient population, we ran additional MCS with a manually increased BSV in total clearance. The coefficient of variation (%CV) (data are median [10–90% percentile]) of 38 β-lactam datasets in CF patients reviewed by Touw et al. (1998) was 27% (16–53%) for total clearance, 29% (17–53%) for volume of distribution at steady state, and 22% (11–45%) for terminal halflife. The %CV in total clearance from our final estimates was approximately 16%. We subsequently increased the % CV in renal clearance to 34% and 65% to achieve variability in total clearance of approximately 27% and 53%, respectively. The latter 2 estimates are the median and 90% percentile of the distribution of variability in total clearance from the 38 β-lactam datasets in CF patients reviewed by Touw et al. (1998).
134
J.B. Bulitta et al. / Diagnostic Microbiology and Infectious Disease 65 (2009) 130–141
3. Results Our CF patients and healthy volunteers had comparable demographic characteristics (Table 1). The noncompartmental parameter estimates shown in Table 2 are not scaled by any size descriptor. CF patients had a 16% higher unscaled total clearance compared with healthy volunteers, which was primarily caused by a 24% higher unscaled renal clearance. The unscaled volume of distribution at steady state was similar between both subject groups. 3.1. Population PK analysis The visual predictive check indicated a better predictive performance for the 3-compartment model than for the 2compartment model. The 1-compartment model had insufficient predictive performance. Fig. 1 reveals the excellent predictive performance of the 3-compartment model with LBM as size descriptor for plasma concentrations and amounts excreted in urine. The visual predictive check showed that the model was able to well capture the central tendency and variability of the plasma and urine profiles for CF patients and healthy volunteers. NONMEM's objective function favored the 3-compartment model by 164 points relative to the 2-compartment model. Therefore, we chose the 3-compartment model as final structural model. The estimates and bootstrap confidence intervals for our final population PK model based on LBM are shown in Table 3. We studied 1) which body size model best described the difference in average clearance and volume of distribution between CF patients and healthy volunteers and 2) which body size model reduced the unexplained BSV most. The ability of the body size models to describe the difference in average PK parameters between both patient groups is shown in Table 4. The scale factors FCYFCLR, FCYFCLNR, and FCYFVSS are the ratios of group estimates between CF patients and healthy volunteers after accounting for body size and renal function. A value of 1.0 for FCYF means that CF patients and healthy volunteers have the same group estimate Table 1 Demographic data (median [minimum–maximum])
No. of subjects (males/females) Height (cm) Body surface areaa (m2) Age (year) WT (kg) LBMb (kg) Body mass index (kg m−2) Serum creatinine concentration (μmol L−1) Renal functionc (mL min−1) a
CF patients
Healthy volunteers
10 (9/1) 178 (167–183) 1.64 (1.48–1.76) 21 (19–25) 54 (47–61) 47.9 (41.9–52.9) 18.0 (14.5–19.2) 64 (49–80)
18 (12/6) 174 (160–203) 1.73 (1.52–2.46) 25 (21–37) 61 (50–107) 49.7 (40.0–82.2) 20.6 (18.1–26.0) 80 (71–95)
162 (132–193)
115 (97–133)
Calculated by the formula from Mosteller (1987). b Calculated by the formula from Cheymol (2000) and James (1976). c Estimated renal function for a subject with nominal WT of 70 kg (see Materials and methods for details).
Table 2 Unscaled PK parameter estimates from noncompartmental analysis for CF patients and healthy volunteers (median [minimum–maximum])
Total clearance (L/h) Renal clearance (L/h) Nonrenal clearance (L/h) Volume of distribution at steady state (L) Fraction excreted unchanged in urine (%) Peak concentration (mg/L) Terminal half-life (L) Mean residence time (h)
CF patients
Healthy volunteers
7.00 (6.14–8.67) 5.72 (3.08–7.19) 1.55 (0.04–3.64) 12.8 (8.01–15.5)
6.05 (4.94–7.46) 4.62 (2.40–5.93) 1.39 (0.63–2.91) 12.1 (9.65–18.4)
80 (46–99) 207 (174–472) 2.00 (1.43–3.10) 1.88 (1.13–2.19)
76 (45–89) 248 (153–352) 2.02 (1.50–2.61) 2.02 (1.65–2.47)
for the respective PK parameter. A value above (below) 1.0 indicates that CF patients have a higher (lower) estimate for the respective PK parameter compared with healthy volunteers with the same body size and same renal function. It is important to note that the estimate for FCYFCLR was derived for a model that included renal function as a covariate on renal clearance. As shown in Table 1, CF patients had a 41% higher average estimated renal function for a nominal 70-kg subject. The estimates for FCYFCLR (Table 4) were close to 1.0 (range, 0.92–1.07 for the different body size models). Thus, CF patients had both an increased creatinine clearance (renal function) and a higher renal clearance of carumonam compared with healthy volunteers. Linear and allometric scaling by WT predicted total clearance and volume of distribution to be 7% to 11% higher in CF patients relative to healthy volunteers (Table 4A). Linear or allometric scaling by LBM explained the difference in average total clearance and volume of distribution between CF patients and healthy volunteers (range, 1.01–1.04 for FCYFCLT and FCYFVSS) (Table 4A). The bootstrap results were consistent with the results for the original dataset. Total clearance was significantly larger in CF patients compared with healthy volunteers for linear scaling by WT because the 90% confidence interval of FCYFCLT did not include 100% for linear scaling by WT (Table 4B). Table 5 shows the ability of the studied body size models to reduce the unexplained BSV. Linear scaling by WT was used as reference model because doses of CF patients are most commonly calculated on a milligram-per-kilogram WT basis. Allometric scaling by WT reduced the unexplained BSV in renal clearance by 38% (P = 0.027 from bootstrap, 1-sided test), and allometric scaling by LBM reduced the unexplained BSV by 30% (P = 0.135), both compared with linear scaling by WT. As final covariate model, we selected allometric scaling by LBM because this model explained the differences in average clearance between both subject groups and reduced the unexplained BSV similarly to allometric scaling by WT. 3.2. Monte Carlo simulation The PTAs were lower for CF patients compared with healthy volunteers (Fig. 2). The following PK–PD MIC
J.B. Bulitta et al. / Diagnostic Microbiology and Infectious Disease 65 (2009) 130–141
135
Fig. 1. Visual predictive check based on 5000 CF patients and 9000 healthy volunteers for plasma concentrations and amounts excreted unchanged in urine for the 3-compartment model based on LBM (Table 3). The plots show the observations, the 80% prediction interval (10–90% percentile), and the interquartile range (25–75% percentile). Ideally, 50% of the observations should fall inside the interquartile range at each time point, and 80% of the observations should fall inside the 80% prediction interval.
breakpoints represent the highest MIC for which the PTA was at least 90% for the target f T N MIC ≥ 60%. Both subject groups had a PK–PD MIC breakpoint of 16 mg/L for continuous infusion at a daily carumonam dose of 6 g per 70 kg of WT at steady state (Fig. 2E and F). For prolonged (4-h) infusions of 2 g per 70 kg of WT q8h, the PK–PD MIC breakpoint was 16 mg/L in healthy volunteers and 8 to 12
mg/L (PTA = 87% for an MIC of 12 mg/L) in CF patients (Fig. 2C and D). For short-term (15-min) infusions of 2 g per 70 kg of WT q8h, the PK–PD MIC breakpoint was 6 mg/L in healthy volunteers and 3 mg/L in CF patients (Fig. 2A and B). This 2-fold difference in PK–PD MIC breakpoints for short-term intermittent treatment seems small. However, the dose in milligram-per-kilogram WT would have to be
136
J.B. Bulitta et al. / Diagnostic Microbiology and Infectious Disease 65 (2009) 130–141
Table 3 PK parameter estimates for the allometric body size model based on allometric scaling by LBM Parameter
Unit
Estimated for
CLTb CLR CLNR Vssb V1 V2 V3 CLicshallow CLicdeep TK0 (fixed) CVC SDC CVAU SDAU
L h−1 L h−1 L h−1 L L L L L h−1 L h−1 min % mg/L % mg
Medianf (90% confidence interval from nonparametric bootstrap)
Estimates from original dataset
CF patients
Healthy volunteers
6.33a,f 4.36 1.97 13.2 7.14 3.70 2.38 14.9 1.55 15 8.7 0.26 38 0.26
6.25f 4.54 1.71 12.7 6.87 3.56 2.29
Estimates for
Coefficient of variation (%)c
12 39 36d 18d 16d
16
CF patients
Healthy volunteers
6.34f (5.90–6.82) 4.38 (4.04–4.78) 1.97 (1.43–2.50) 12.8 (11.6–14.1) 6.88 (5.64–8.27) 3.42 (2.89–3.92) 2.45 (2.06–2.93) 14.3 (9.94–18.6) 1.58 (1.23–2.16) 15 8.7 (7.4–10) 0.23 (0.11–0.55) 36 (27–47) 0.26 (0.0026–1.18)
6.26f (5.92–6.63) 4.51 (4.19–4.86) 1.75 (1.46–2.05) 12.2 (11.5–13.0) 6.64 (5.56–7.59) 3.28 (2.77–3.76) 2.34 (1.97–2.81)
Coefficient of variation (%)c
11 (7.8–15) 36 (22–48) 34e (21–48) 21e (14–30) 16e (8.6–22)
16 (9.7–23)
CLT = total clearance; V1 = volume of distribution for the central compartment; V2 = volume of distribution for the shallow peripheral compartment; V3 = volume of distribution for the deep peripheral compartment; Vss = volume of distribution at steady state; CLicshallow = intercompartmental clearance between the central and the shallow peripheral compartment; CLicdeep = intercompartmental clearance between the central and the deep peripheral compartment; TK0 = duration of zero-order input (not estimated). CVC is the proportional and SDC is the additive residual error component for the plasma concentrations. CVAU is the proportional and SDAU is the additive residual error component for the amounts excreted in urine. a All clearance and volume terms are group estimates for subjects of standard body size (LBMSTD = 53 kg). b Derived from model estimates, not an estimated parameter. c Apparent coefficient of variation for the BSV. d The coefficients of correlation for pairs of random effects were rBSV(V1,V2) = −0.61, rBSV(V1,V3) = −0.22, and rBSV(V2,V3) = 0.06. e Median (90% confidence intervals) from bootstrap: rBSV(V1,V2) = −0.74 (−0.91 to −0.30), rBSV(V1,V3) = −0.11 (−0.52 to 0.31), and rBSV(V2,V3) = −0.048 (−0.55 to 0.38). f For the analysis of the original dataset, the estimates of structural model parameters are geometric means. For the bootstrap analysis, these estimates are the median of 1000 geometric means of the respective parameter.
doubled in CF patients, if CF patients should achieve the same PTAs as healthy volunteers in empiric therapy. We additionally explored dosing algorithms that aimed at achieving the same f T N MIC in CF patients of various body
sizes. These optimized dosing algorithms were based on an allometric scaling by LBM or WT and increased the PK–PD MIC breakpoints by about 5% to 15% compared with milligram-per-kilogram WT dosing for the 3 studied
Table 4 Ratio of group estimates (CF patients/healthy volunteers) for clearance and volume of distribution for different body size models Part A
Estimates from original dataset
Size model
FCYFCLR
FCYFCLNR
FCYFCLTa
FCYFVSS
1) 2) 3) 4) 5)
0.92 1.06 1.02 0.99 0.96
1.02 1.25 1.19 1.13 1.15
0.95 1.11 1.07 1.03 1.01
0.99 1.10 1.11 1.04 1.04
No size model WT linear scaling WT allometric LBM linear scaling LBM allometric
Part B
Median (90% confidence interval) from bootstrap replicates
Size model
FCYFCLR
FCYFCLNR
FCYFCLTa
FCYFVSS
1) 2) 3) 4) 5)
0.92 (0.84–1.00) 1.07 (0.95–1.22) 1.02 (0.92–1.15) 1.00 (0.89–1.13) 0.97 (0.87–1.09)
1.02 (0.75–1.39) 1.24 (0.88–1.66) 1.18 (0.86–1.59) 1.13 (0.82–1.50) 1.13 (0.79–1.49)
0.95 (0.87–1.04) 1.12 (1.02–1.23) 1.07 (0.98–1.17) 1.04 (0.94–1.14) 1.01 (0.93–1.10)
0.97 (0.85–1.09) 1.11 (1.00–1.26) 1.12 (1.00–1.27) 1.04 (0.95–1.14) 1.04 (0.95–1.15)
No size model WT linear scaling WT allometric LBM linear scaling LBM allometric
FCYFNNN = ratio of group estimates for parameter NNN after adjusting for body size and renal function (group estimate for CF patients divided by group estimate for healthy volunteers). a Calculated as FCYFCLT = (FCYFCLR · CLPOP,R + FCYFCLNR · CLPOP,NR)/(CLPOP,R + CLPOP,NR); the group estimates for renal and nonrenal clearance of healthy volunteers with standard body size are denoted as CLPOP,R and CLPOP,NR.
J.B. Bulitta et al. / Diagnostic Microbiology and Infectious Disease 65 (2009) 130–141 Table 5 Comparison of BSV (variances) between the different body size models Size model
Estimates from original dataset (%) CLR
2) WT linear scaling 3) WT allometric 4) LBM linear scaling 5) LBM allometric
a
100 62b,c 99c 70c
CLNR a
100 94 102 103
V1
V2 a
100 98 99 115
V3 a
100 99 85 73
100a 101 78 70
The table shows the variance for each body size model divided by the variance for linear scaling by WT for the respective PK parameter (Table 3 for parameter explanations). a The between-subject variances were scaled to the variance for linear scaling by WT. b Significantly smaller than 100% (P = 0.027 from 1000 bootstrap, 1sided test). All other 90% confidence intervals included 100%. c The lower this number, the more variability was explained by the respective body size model. These values mean that the BSV (variance) for renal clearance was reduced by 38% for allometric scaling by WT, by 1% for linear scaling by LBM, and by 30% for allometric scaling by LBM, all compared with linear scaling by WT.
carumonam dosage regimens (data not shown). Based on our population PK parameter estimates for healthy volunteers (Table 3), the expected PTA was 88% for continuous infusion, 86% for 4-h infusion q8h, and 65% for 15-min infusion for the MIC distribution in Japanese patients with urinary tract infections by P. aeruginosa at a daily dose of 6 g per 70 kg of WT. In MCS with larger variability in total clearance, the PK– PD MIC breakpoints for continuous and 4-h infusions q8h were approximately 1.5-fold lower for a %CV of 27% for total clearance and 2-fold lower for a %CV of 53% for total clearance compared with the breakpoints based on the final model (Table 2). For 15-min infusions q8h, PK–PD MIC breakpoints were 2-fold lower for a %CV of 27% for total clearance and 4-fold lower for a %CV of 53% for total clearance.
4. Discussion The improved life expectancy of CF patients has led to a significantly increased use of antibiotics in adult CF patients. However, only a few studies systematically compared the PK of antibiotics in CF patients and healthy volunteers. The ratio of mean clearance (in liters per hour per kilogram or liters per hour per 1.73 m2 body surface area) in CF patients divided by mean clearance in healthy volunteers ranges from 95% to 254% for β-lactams (Touw et al., 1998). This range is 64% to 144% for volume of distribution (in liters per kilogram or liters per 1.73 m2). A better understanding of these, in part, pronounced differences is important to adequately scale PK parameters by body size and optimize dosage regimens for CF patients of various body sizes. We studied various descriptors for body size (functional capacity) with linear or allometric scaling. A priori, we believed that it is important to account for differences in
137
body composition. Our study in adult subjects was rather small and had a relatively narrow range in body size. We incorporated body size and renal function as covariates based on prior knowledge, because covariate model building without prior knowledge is difficult for studies with small sample size (Ribbing and Jonsson, 2004). Carumonam was expected to primarily distribute into aqueous body compartments because β-lactams are hydrophilic molecules. Our CF patients and healthy volunteers had a similar LBM (Table 1). Thus, both groups were likely to have a comparable amount of aqueous body space and had a comparable volume of distribution (Table 2). Our PK parameter estimates were in good agreement with reports from other authors (Berube et al., 1988; Koeppe et al., 1987) for healthy volunteers. Unscaled renal clearance was 24% higher in CF patients (Table 2), which corresponded to the higher renal function in our CF patients compared with our healthy volunteers. This result is in agreement with studies in CF patients and healthy volunteers on other β-lactams (Bulitta et al., 2007; Hedman et al., 1988, 1990; Vinks et al., 2007; Wang et al., 1993). Renal function as a covariate on renal clearance explained the difference in average renal clearance of carumonam between CF patients and healthy volunteers and reduced the unexplained variance in renal clearance by 31% (from 14% to 12% CV). Therefore, renal function was a useful covariate in our study. However, our study was relatively small and is limited in representing the whole population of CF patients. In addition, Al-Aloul et al. (2007) pointed out limitations of the Cockcroft and Gault formula for predicting creatinine clearance in CF patients. Because our CF patients had a similar LBM compared with our healthy volunteers, the scale factor for clearance and volume of distribution was close to 1.0 (Table 4). CF patients had higher clearances and a larger volume of distribution for linear and allometric scaling by WT, probably because WT does not account for differences in body composition. Linear or allometric scaling by LBM explained the difference in average clearance and volume of distribution between both subject groups (Table 4). Allometric scaling by WT or LBM reduced the unexplained BSV by 30% to 38% for renal clearance, and allometric scaling by LBM reduced the unexplained BSV for volume of distribution of the peripheral compartments by about 30% (Table 5). Although renal function and LBM explained the difference in average PK parameters between both subject groups, CF patients still had a higher renal clearance than healthy volunteers. In agreement with the results for aztreonam from Vinks et al. (2007), CF patients achieved about 2 times lower PK–PD MIC breakpoints compared with healthy volunteers for intermittent short-term infusion with carumonam and dosing on a milligram-per-kilogram WT basis (Fig. 2). The lower PK–PD MIC breakpoints in CF patients were primarily a consequence 1) of the slightly higher renal clearance (and renal function) and 2) of the altered body composition of CF patients. It depends on the MIC distribution in the target patient population, whether or not a 2-fold difference in
138
J.B. Bulitta et al. / Diagnostic Microbiology and Infectious Disease 65 (2009) 130–141
Fig. 2. PTA for a daily carumonam dose of 6 g per 70 kg of WT at steady state for different targets of the PK–PD index fT N MIC.
PK–PD MIC breakpoints has a pronounced impact on the probability of successful treatment. Instead of increasing the dose, we studied prolonged and continuous infusion as alternative modes of administration to increase PK–PD MIC breakpoints. For the target f T N MIC ≥ 60%, we found a PK–PD MIC breakpoint of 8 to 12 mg/L (PTA = 87% at an MIC of 12 mg/L) in CF patients and of 16 mg/L in healthy volunteers for prolonged (4-h) infusions of 2 g per 70 kg of WT q8h (Fig. 2C and D). Continuous infusion at the same daily dose of 6 g per 70 kg of WT achieved a PK– PD MIC breakpoint of 16 mg/L in both subject groups. Dosing algorithms based on allometric scaling by LBM (Bulitta et al., 2006) yielded an additional improvement in the PK–PD MIC breakpoint by about 5% to 15% for the studied short-term, prolonged, and continuous infusion regimens.
Our estimates for the variability in renal and nonrenal clearance (Table 2) resulted in a %CV of 16% for total clearance. Because the variability of total clearance for βlactams in studies with CF patients (Touw et al., 1998) was larger than in our well-controlled study, we ran MCS with a %CV of 27% and 53% in total clearance. These are the median and 90% percentile of the variability in total clearance of 38 datasets on β-lactams (Touw et al., 1998). As expected, the PK–PD MIC breakpoints were lower (up to 4-fold for 15-min infusions q8h) compared with the MCS with the 16% CV in total clearance. To put these PTAs into clinical perspective, the expected PTA for treatment of infections caused by bacteria from a specific MIC distribution needs to be calculated (e.g., for the MIC distribution of a local hospital). Prolonged and
J.B. Bulitta et al. / Diagnostic Microbiology and Infectious Disease 65 (2009) 130–141
continuous infusions are expected to achieve notably higher PTAs than short-term infusions for some MICs. It is important to determine these MICs. Carumonam has good activity against Proteus mirabilis, K. pneumoniae, S. marcescens, Serratia liquefaciens, and E. coli with MIC90 values between ≤0.125 and 0.5 mg/L (Igari et al., 2003; Ikemoto et al., 1996a; Kumamoto et al., 2001, 2004, 2006). If the MIC90 is below 1 mg/L, short-term intermittent infusions at a daily dose of 6 g per 70 kg of WT were predicted to achieve a PTA expectation value N95%. However, other authors report a considerably lower susceptibility to carumonam. S. marcescens has been reported to have an MIC90 of 16 mg/L (Kumamoto et al., 2006), and P. aeruginosa had an MIC80 of 2 mg/L (Ikemoto et al., 1996b, 1997). More recent data on P. aeruginosa show an MIC90 of 32 mg/L, with 61% of the isolates having an MIC of 2 to 16 mg/L (Kumamoto et al., 2006). It is the infections caused by these pathogens for which prolonged and continuous infusion will have considerably higher PTA expectation values compared with short-term infusion. This applies especially to CF patients because they had a breakpoint of 3 mg/L for short-term infusions of 2 g per 70 kg of WT q8h and a breakpoint of 8 to 12 mg/L for prolonged infusions at the same dose. Two limitations of our MCS for CF patients are that we did not measure concentrations of carumonam in the lungs of CF patients and that we are not aware of an established PK– PD target for successful therapy in CF patients. Especially for CF patients with chronic infections by P. aeruginosa, it seems likely that substantially higher PK–PD targets and possibly also combination therapy are required for successful treatment compared with early P. aeruginosa infections (Gibson et al., 2003). In conclusion, we found a 24% higher unscaled renal clearance in CF patients compared with healthy volunteers that could be explained by an enhanced glomerular filtration rate in CF patients. The difference in average PK parameters between CF patients and healthy volunteers could be explained by body size, body composition, and renal function. These differences were better explained by LBM than by WT. Allometric scaling by LBM reduced the unexplained BSV in renal clearance and volume of distribution of both peripheral compartments by about 30% relative to linear scaling by WT. Future clinical studies are warranted to explore the expected higher probability of successful clinical outcome for allometric dose selection in CF patients based on LBM. Prolonged and continuous infusion achieved higher PK–PD MIC breakpoints compared with short-term infusions q8h at the same daily dose. At a daily dose of 6 g per 70 kg of WT, CF patients achieved a PK–PD MIC breakpoint (target: f T N MIC ≥ 60%) of 3 mg/L for 15-min infusions q8h, of 8 to 12 mg/L (87% PTA for MIC = 12 mg/L) for 4-h infusions q8h, and of 16 mg/L for continuous infusion. Therefore, prolonged infusion was predicted to be especially superior to short-term infusion, if the MIC50 and MIC90 fall between 2 and 8 mg/L.
139
Acknowledgment The authors thank Dr Nick Holford and Dr Thomas Lodise for very fruitful discussions and comments on this project and manuscript. References Al-Aloul M, Jackson M, Bell G, Ledson M, Walshaw M (2007) Comparison of methods of assessment of renal function in cystic fibrosis (CF) patients. J Cyst Fibros 6:41–47. Anderson BJ, Holford NH (2008) Mechanism-based concepts of size and maturity in pharmacokinetics. Annu Rev Pharmacol Toxicol 48: 303–332. Beal SL, Boeckmann AJ, Sheiner LB, NONMEM Project Group (1999) NONMEM users guides. San Francisco: University of California at San Francisco. Berube D, Vallee F, Panneton AC, Bergeron MG, LeBel M (1988) Pharmacokinetics of carumonam after single and multiple 1- and 2-g dosage regimens. Antimicrob Agents Chemother 32:354–357. Bulitta JB, Duffull SB, Landersdorfer CB, Drusano GL, Kinzig-Schippers M, Holzgrabe U, Stephan U, Sörgel F (2006) Pharmacodynamic comparison of cystic fibrosis patients and healthy volunteers by population pharmacokinetics (PK) and Monte Carlo simulation. Poster W4056. AAPS Annual Meeting, San Antonio, TX, USA. Bulitta JB, Duffull SB, Kinzig-Schippers M, Holzgrabe U, Stephan U, Drusano GL, Sorgel F (2007) Systematic comparison of the population pharmacokinetics and pharmacodynamics of piperacillin in cystic fibrosis patients and healthy volunteers. Antimicrob Agents Chemother 51:2497–2507. Cheymol G (2000) Effects of obesity on pharmacokinetics implications for drug therapy. Clin Pharmacokinet 39:215–231. Cockcroft DW, Gault MH (1976) Prediction of creatinine clearance from serum creatinine. Nephron 16:31–41. Craig WA (1998) Pharmacokinetic/pharmacodynamic parameters: rationale for antibacterial dosing of mice and men. Clin Infect Dis 26:1–12. Craig WA (2003) Basic pharmacodynamics of antibacterials with clinical applications to the use of β-lactams, glycopepq8hes, and linezolid. Infect Dis Clin North Am 17:479–501. Drusano GL (2004) Antimicrobial pharmacodynamics: critical interactions of “bug and drug”. Nat Rev Microbiol 2:289–300. Efron B, Tibshirani RJ (1986) Bootstrap methods for standard errors, confidence intervals and other measures of statistical accuracy. Stat Sci 1:54–77. Gibson RL, Burns JL, Ramsey BW (2003) Pathophysiology and management of pulmonary infections in cystic fibrosis. Am J Respir Crit Care Med 168:918–951. Gottschalk B, Sorgel F, Wiesemann HG, Stephan U, Rey M, Bowing HB, Stehr C, Dominick HC, Kiosz D, Geldmacher-von Mallinckrodt M (1988) Antibiotic pharmacokinetics in patients with mucoviscidosis. Monatsschr Kinderheilkd 136:475–478. Green B, Duffull SB (2004) What is the best size descriptor to use for pharmacokinetic studies in the obese? Br J Clin Pharmacol 58:119–133. Hedman A, Adan-Abdi Y, Alvan G, Strandvik B, Arvidsson A (1988) Influence of the glomerular filtration rate on renal clearance of ceftazidime in cystic fibrosis. Clin Pharmacokinet 15:57–65. Hedman A, Alvan G, Strandvik B, Arvidsson A (1990) Increased renal clearance of cefsulodin due to higher glomerular filtration rate in cystic fibrosis. Clin Pharmacokinet 18:168–175. Holford NH (1996) A size standard for pharmacokinetics. Clin Pharmacokinet 30:329–332. Igari J, Oguri T, Hiramatsu N, Akiyama K, Koyama T (2003) Postmarketing surveillance of antibacterial activities of cefozopran against various clinical isolates—II. Gram-negative bacteria. Jpn J Antibiot 56: 458–496.
140
J.B. Bulitta et al. / Diagnostic Microbiology and Infectious Disease 65 (2009) 130–141
Ikemoto H, Watanabe K, Mori T, Igari J, Oguri T, Kobayashi K, Satou K, Matsumiya H, Saito A, Terai T, Tanno Y, Nishioka K, Arakawa M, Wada K, Okada M, Ozaki K, Aoki N, Kitamura N, Sekine O, Suzuki Y, Matsuda M, Tanimoto H, Nakata K, Nakamori Y, Kusano N, et al. (1996a) Susceptibilities of bacteria isolated from patients with respiratory infectious diseases to antibiotics (1994). Jpn J Antibiot 49:419–455. Ikemoto H, Watanabe K, Mori T, Igari J, Oguri T, Kobayashi K, Satou K, Matsumiya H, Saito A, Terai T, Tanno Y, Nishioka K, Arakawa M, Wada K, Okada M, Ozaki K, Aoki N, Kitamura N, Sekine O, Suzuki Y, Tanimoto H, Nakata K, Nakamori Y, Nakatani T, Kusano N, et al. (1996b) Susceptibilities of bacteria isolated from patients with respiratory infectious diseases to antibiotics (1992). Jpn J Antibiot 49:34–70. Ikemoto H, Watanabe K, Mori T, Igari J, Oguri T, Kawaguchi H, Shimizu Y, Matsumiya H, Saito A, Terai T, Inoue H, Nakadate T, Ito C, Yosida T, Tanno Y, Ohno I, Nishioka K, Arakawa M, Igarashi K, Wada K, Okada M, Ozaki K, Aoki N, Kitamura N, Touyama M, et al. (1997) Susceptibilities of bacteria isolated from patients with respiratory infectious diseases to antibiotics (1995). Jpn J Antibiot 50:421–459. Ikemoto H, Watanabe K, Mori T, Igari J, Oguri T, Shimizu Y, Terai T, Inoue H, Nakadate T, Ito C, Yoshida T, Ohno I, Tanno Y, Arakawa M, Igarashi K, Okada M, Ozaki K, Aoki N, Kitamura N, Sekine O, Suzuki Y, Nakata K, Nakatani T, Inagawa H, Kusano N (1998) Susceptibilities of bacteria isolated from patients with lower respiratory infectious diseases to antibiotics (1996). Jpn J Antibiot 51:437–474. Imada A, Kondo M, Okonogi K, Yukishige K, Kuno M (1985) In vitro and in vivo antibacterial activities of carumonam (AMA-1080), a new N-sulfonated monocyclic beta-lactam antibiotic. Antimicrob Agents Chemother 27:821–827. James W (1976) Research on Obesity. London: Her Majesty's Stationery Office. Jones RN, Barry AL, Thornsberry C, Fuchs PC, Packer RR (1986) The antimicrobial activity, beta-lactamase stability, and disk diffusion susceptibility testing of carumonam (RO 17-2301, AMA-1080), a new monobactam. Am J Clin Pathol 86:608–618. Koeppe P, Hoffler D, Strobel K (1987) Pharmacokinetics and dose recommendations of carumonam in renal failure. Arzneimittelforschung 37:65–69. Kumamoto Y, Tsukamoto T, Watanabe K, Kobayashi Y, Hirose T, Matsukawa M, Uchida H, Takahashi S, Kunishima Y, Matsuda S, Sato S, Ogihara M, Ishibashi K, Shigeta S, Fujime M, Fujita K, Yoshida H, Imafuku Y, Igari J, Murai M, Oguri T, Yamaguchi K, Matsumoto T, Kashitani F, Furuhama T, Kumon H, Monden K, Ooe H, Nishikawa M, Kohno S, Miyazaki Y, Izumikawa K, Yamaguchi T, Mochida C, Oka T, Kitamura M, Takano Y, Matsuoka Y (2001) Comparative studies on activities of antimicrobial agents against causative organisms isolated from patients with urinary tract infections (1999). I. Susceptibility distribution. Jpn J Antibiot 54:185–216. Kumamoto Y, Tsukamoto T, Watanabe K, Kobayashi Y, Matsukawa M, Uchida H, Kunishima Y, Hirose T, Matsuda S, Sato S, Shigeta S, Fujime M, Fujita K, Yamaguti O, Ogihara M, Ishibashi K, Igari J, Takahashi K, Oguri T, Yoshida H, Imafuku Y, Yamaguchi K, Furuya N, Kashitani F, Murai M, Ooe H, Nishikawa M, Kumon H, Monden K, Oka T, Kitamura M, Kohno S, Tomono K, Miyazaki Y, Matsuoka Y, Fukuhara Y, Hirakata Y, Aoki S (2003) Comparative studies on activities of antimicrobial agents against causative organisms isolated from patients with urinary tract infections (2001). I. Susceptibility distribution. Jpn J Antibiot 56: 396–423. Kumamoto Y, Tsukamoto T, Matsukawa M, Kunishima Y, Hirose T, Yamaguti O, Ishibashi K, Shigeta S, Suzutani T, Yoshida H, Imafuku Y, Murai M, Watanabe K, Kobayashi Y, Uchida H, Matsuda S, Sato S, Fujime M, Fujita K, Igari J, Oguri T, Yamaguchi K, Furuya N, Deguchi T, Ishihara S, Ooe H, Nishikawa M, Oka T, Kitamura M, Fukuhara Y, Kamidono S, Arakawa S, Kumon H, Monden K, Matsumoto T, Takahashi K, Naito S, Egashira T, Kohno S, Miyazaki Y, Hirakata Y, Aoki S (2004) Comparative studies on activities of antimicrobial agents against causative organisms isolated from patients with urinary tract infections (2002). I. Susceptibility distribution. Jpn J Antibiot 57:246–274.
Kumamoto Y, Tsukamoto T, Matsukawa M, Kunishima Y, Hirose T, Shigeta S, Yamaguti O, Ishibashi K, Suzutani T, Yoshida H, Imafuku Y, Murai M, Watanabe K, Kobayashi Y, Uchida H, Matsuda S, Sato S, Fujime M, Fujita K, Igari J, Oguri T, Yamaguchi K, Furuya N, Deguchi T, Ishihara S, Ooe H, Oka T, Kitamura M, Fukuhara Y, Kamidono S, Arakawa S, Kumon H, Monden K, Matsumoto T, Muratani T, Naito S, Egashira T, Konishi T, Kohno S, Hirakata Y, Kondo A, Matsuda J, Nakano M (2006) Comparative studies on activities of antimicrobial agents against causative organisms isolated from patients with urinary tract infections (2004). I. Susceptibility distribution. Jpn J Antibiot 59: 177–200. McNulty CA, Garden GM, Ashby J, Wise R (1985) Pharmacokinetics and tissue penetration of carumonam, a new synthetic monobactam. Antimicrob Agents Chemother 28:425–427. Miller ME, Kornhauser DM (1994) Bromide pharmacokinetics in cystic fibrosis. Arch Pediatr Adolesc Med 148:266–271. Morgan DJ, Bray KM (1994) Lean body mass as a predictor of drug dosage. Implications for drug therapy. Clin Pharmacokinet 26:292–307. Mosteller RD (1987) Simplified calculation of body-surface area. N Engl J Med 317:1098. Paap CM, Nahata MC (1990) Clinical pharmacokinetics of antibacterial drugs in neonates. Clin Pharmacokinet 19:280–318. Parke J, Holford NH, Charles BG (1999) A procedure for generating bootstrap samples for the validation of nonlinear mixed-effects population models. Comput Methods Programs Biomed 59:19–29. Prandota J (1987) Drug disposition in cystic fibrosis: progress in understanding pathophysiology and pharmacokinetics. Pediatr Infect Dis J 6:1111–1126. Prandota J (1988) Clinical pharmacology of antibiotics and other drugs in cystic fibrosis. Drugs 35:542–578. Raimondi A, Mattina R, Cocuzza CE (1988) Carumonam's in-vitro activity against gram-negative bacteria and its stability to their beta-lactamases. Chemioterapia 7:3–9. Rey E, Treluyer JM, Pons G (1998) Drug disposition in cystic fibrosis. Clin Pharmacokinet 35:313–329. Ribbing J, Jonsson EN (2004) Power, selection bias and predictive performance of the Population Pharmacokinetic Covariate Model. J Pharmacokinet Pharmacodyn 31:109–134. Rochat T, Slosman DO, Pichard C, Belli DC (1994) Body composition analysis by dual-energy x-ray absorptiometry in adults with cystic fibrosis. Chest 106:800–805. Shinagawa N, Yura J, Manabe T, Mashita K, Ishikawa S, Mizuno A, Hirata K, Denno R, Mukaiya M, Ishibiki K, Ushijima Y, Aikawa N, Takuma K, Iwai S, Kunimatsu M, Ohtsuka K, Kinoshita H, Morimoto K, Fujimoto M, Tanimura H, Ohnishi H, Umemoto Y, Sakaguchi S, Dounishi H, Oda S, et al. (1996) Isolation rate of Pseudomonas aeruginosa from surgical infections and their susceptibilities. Jpn J Antibiot 49:544–554. Slosman DO, Casez JP, Pichard C, Rochat T, Fery F, Rizzoli R, Bonjour JP, Morabia A, Donath A (1992) Assessment of whole-body composition with dual-energy x-ray absorptiometry. Radiology 185:593–598. Sorgel F, Stephan U, Wiesemann HG, Gottschalk B, Stehr C, Rey M, Bowing HB, Dominick HC, Geldmacher von Mallinckrodt M (1987) High dose treatment with antibiotics in cystic fibrosis—a reappraisal with special reference to the pharmacokinetics of β-lactams and new fluoroquinolones in adult CF-patients. Infection 15:385–396. Sweetman SC (2009) Martindale: the complete drug reference [online]. London: Pharmaceutical Press. pp. 36. http://www.medicinescomplete. com/. Touw DJ (1998) Clinical pharmacokinetics of antimicrobial drugs in cystic fibrosis. Pharm World Sci 20:149–160. Touw DJ, Vinks AA, Heijerman HG, Hermans J, Bakker W (1994) Suggestions for the optimization of the initial tobramycin dose in adolescent and adult patients with cystic fibrosis. Ther Drug Monit 16: 125–131. Touw DJ, Vinks AA, Heijerman HG, Bakker W (1996) Prospective evaluation of a dose prediction algorithm for intravenous tobramycin in
J.B. Bulitta et al. / Diagnostic Microbiology and Infectious Disease 65 (2009) 130–141 adolescent and adult patients with cystic fibrosis. Ther Drug Monit 18: 118–123. Touw DJ, Vinks AA, Mouton JW, Horrevorts AM (1998) Pharmacokinetic optimisation of antibacterial treatment in patients with cystic fibrosis. Current practice and suggestions for future directions. Clin Pharmacokinet 35:437–459. Vinks AA, van Rossem RN, Mathot RA, Heijerman HG, Mouton JW (2007) Pharmacokinetics of aztreonam in healthy subjects and patients with cystic fibrosis and evaluation of dose–exposure relationships using Monte Carlo simulation. Antimicrob Agents Chemother 51: 3049–3055. Wang JP, Unadkat JD, al-Habet SM, O'Sullivan TA, Williams-Warren J, Smith AL, Ramsey B (1993) Disposition of drugs in cystic fibrosis. IV.
141
Mechanisms for enhanced renal clearance of ticarcillin. Clin Pharmacol Ther 54:293–302. West GB, Brown JH, Enquist BJ (1997) A general model for the origin of allometric scaling laws in biology. Science 276:122–126. West GB, Brown JH, Enquist BJ (1999) The fourth dimension of life: fractal geometry and allometric scaling of organisms. Science 284:1677–1679. Yoshida K, Mitani M, Naeshiro I, Torii H, Tanayama S (1986) Disposition of carumonam (AMA-1080/Ro 17-2301), a new N-sulfonated monocyclic β-lactam, in rats and dogs. Antimicrob Agents Chemother 29: 1017–1024. Zhao WH, Hu ZQ, Shimura T (2008) Potency of IMP-10 metallo-betalactamase in hydrolysing various antipseudomonal beta-lactams. J Med Microbiol 57:974–979.