J. vet. Pharmacol. Therap. doi: 10.1111/jvp.12156
Population pharmacokinetics of rifampin in the treatment of Mycobacterium tuberculosis in Asian elephants E. F. EGELUND* R. ISAZA
†
A. P. BROCK †
Egelund, E. F., Isaza, R., Brock, A. P., Alsultan, A., An, G., Peloquin, C. A. Population pharmacokinetics of rifampin in the treatment of Mycobacterium tuberculosis in Asian elephants. J. vet. Pharmacol. Therap. doi: 10.1111/jvp. 12156.
A. ALSULTAN* G. AN ‡ & C. A. PELOQUIN* *Department of Pharmacotherapy and Translational Research, College of Pharmacy, University of Florida, Gainesville, FL, USA; † Department of Small Animal Clinical Sciences, College of Veterinary Medicine, University of Florida, Gainesville, FL, USA; ‡ Department of Pharmaceutics, College of Pharmacy, University of Florida, Orlando, FL, USA
The objective of this study was to develop a population pharmacokinetic model for rifampin in elephants. Rifampin concentration data from three sources were pooled to provide a total of 233 oral concentrations from 37 Asian elephants. The population pharmacokinetic models were created using Monolix (version 4.2). Simulations were conducted using ModelRisk. We examined the influence of age, food, sex, and weight as model covariates. We further optimized the dosing of rifampin based upon simulations using the population pharmacokinetic model. Rifampin pharmacokinetics were best described by a one-compartment open model including first-order absorption with a lag time and first-order elimination. Body weight was a significant covariate for volume of distribution, and food intake was a significant covariate for lag time. The median Cmax of 6.07 lg/mL was below the target range of 8–24 lg/mL. Monte Carlo simulations predicted the highest treatable MIC of 0.25 lg/mL with the current initial dosing recommendation of 10 mg/kg, based upon a previously published target AUC0–24/MIC > 271 (fAUC > 41). Simulations from the population model indicate that the current dose of 10 mg/kg may be adequate for MICs up to 0.25 lg/mL. While the targeted AUC/MIC may be adequate for most MICs, the median Cmax for all elephants is below the human and elephant targeted ranges. (Paper received 21 May 2014; accepted for publication 10 July 2014) Charles Peloquin, PharmD, FCCP, Professor, and Director, Infectious Disease Pharmacokinetics Laboratory, College of Pharmacy, and Emerging Pathogens Institute, University of Florida, 1600 SW Archer Road, Rm P4-33, PO Box 100486, Gainesville, FL 32610-0486, USA. E-mail: [email protected]
INTRODUCTION Mycobacterium tuberculosis (MTb), the causative agent of tuberculosis (TB), is believed to have affected elephant populations for centuries (McGaughey, 1961). Limited data exist regarding the appropriate treatment of MTb in elephants. Currently, elephants are treated with the same drug regimen used in humans, with doses based on linear scaling of weight (USDA, 2008). Only recently have pharmacokinetic studies been conducted to determine the concentration–time profiles and pharmacokinetic parameters associated with the doses used in the treatment of TB in elephants (Maslow et al., 2005a,b; Zhu et al., 2005; Peloquin et al., 2006; Brock et al., 2014). At the time of this writing, only one previous paper exists which examines the population pharmacokinetics of rifampin in elephants (Peloquin et al., 2006). © 2014 John Wiley & Sons Ltd
Along with isoniazid, rifampin is the most important drug for treating tuberculosis, in both humans and elephants. However, the plasma concentrations required for efficacy in elephants remain unknown. Currently, dose recommendations are based on human plasma concentrations, with a target maximum concentration (Cmax) range of 8–24 lg/mL (Acocella, 1978; USDA, 2008). This target Cmax range lies well above the minimum inhibitory concentration (MIC) range typically seen with MTb (0.15–0.5 lg/mL). Ideally, the initial recommended rifampin dose for elephants is 10 mg/kg, to be followed by therapeutic drug monitoring (TDM) (USDA, 2008). Only sparse concentration data were available previously, making it difficult to assess the current recommended dose in relation to the drug’s Cmax. Several published studies confirm that rifampin’s antibacterial effect is concentration dependent, requiring an optimization of either Cmax or the area under the 1
2 E. F. Egelund et al.
curve (AUC), that is, the drug’s exposure, over the drug’s MIC to maximize the killing of MTb (Verbist, 1969; Jayaram et al., 2003; Nuermberger & Grosset, 2004; Gumbo et al., 2007). Recent publications have cited the AUC/MIC as the pharmacokinetic/pharmacodynamic (PK/PD) parameter most highly correlated with killing, while Cmax/MIC is most associated with prevention of resistance and postantibiotic effect (Jayaram et al., 2003; Gumbo et al., 2007). Additionally, covariates that could influence the between-elephant PK variability have not been identified. A greater understanding of these factors is needed to optimize treatment regimens and prevent resistance (Dumonceaux et al., 2011). Nonlinear mixed effect (NLME) models enable investigators to determine between-subject variability (BSV), accounting for both fixed and random effects. Fixed effects are population parameters which are considered constant (e.g., Vd, Cl) while random effects refer to unexplained variation. NLME allows the utilization of both intensive and sparse data. The aim of this study was to develop a population pharmacokinetic model using available intensive and sparse data to examine the influence of specific covariates on rifampin’s pharmacokinetics. Further, we assessed the current dosing recommendations using simulations based upon our population model.
although some had more than two concentration–time points. Dosing at 10 mg/kg was either rectal or oral. Subset 3 consisted of 214 concentrations from 23 elephants. Elephants were both intensively and sparsely sampled after either rectal (two methods) or oral administration (four methods) with or without food, as previously described (Peloquin et al., 2006). Dosing ranged from 4 to 12 mg/kg. Sample collection, storage, and analysis For each subset, samples were collected from the caudal auricular vein into lithium heparin tubes. Samples then were placed on ice until centrifugation. Plasma was harvested and stored on dry ice prior to transfer for up to 24 h and then stored at 80 °C until assayed. Prior to analysis, samples were allowed to thaw at room temperature. For each subset, concentrations were determined by high-performance liquid chromatography using a validated assay described previously (Peloquin et al., 1999). Briefly, the rifampin standard curve ranged from 0.5 to 50 lg/mL. The lower limit of quantification (LLOQ) was the lowest point of the standard curve (i.e., 0.5 lg/mL). All assays were free of any interfering peaks from other study drugs and certified by the College of American Pathologists. Pharmacokinetic data analysis
MATERIALS AND METHODS Study population Data from three sources were pooled to form the data set used in this analysis. Four hundred and fifteen samples (233 oral, 182 rectal) were collected from 37 elephants, weighing 849– 7345 kg (see Table 1 for full demographics). Subset 1 consisted of 154 concentrations intensively sampled from six elephants. Plasma samples were collected at 14 time points. Rifampin was administered orally at 10 mg/kg to six elephants and rectally at 10 mg/kg to five of these elephants both as a single dose. Elephants were allowed food and water ad libitum (Brock et al., 2014). Subset 2 consisted of 62 concentrations from eight elephants on different occasions. Elephants were sparsely sampled, primarily at 2 and 4 h after drug administration for TDM,
A majority of plasma concentrations following rectal administration were below the limit of quantification (BLQ), thus all rectal concentrations were excluded. A number of concentrations in the oral data set also were BLQ. Potential data handling options for BLQ values include using the raw data, replacing the data with half the BLQ, omitting the BLQ data, or using likelihood-based approaches (Ahn et al., 2008). Each of these methods can introduce bias to the modeling process (Duval & Karlsson, 2002). We elected to include the BLQ data because most were within 10% of the LLOQ. Further, data from each subject were graphed and evaluated to verify that any BLQ data were pharmacokinetically plausible. A compartmental model using NLME modeling was developed to characterize pharmacokinetic parameters. The software, Monolix v 4.2 (Lixoft. Orsay, France), was used to analyze the data. Several structural models were evaluated,
Table 1. Summary of subject demographics (median and range) and rifampin sample data (median and range) for purposes of a population pharmacokinetic model in Asian elephants Characteristics No. of subjects Female sex Median weight in kg (range) Median age (range) No. of samples Oral Rectal Sampling Steady-state Oral median Cmax in lg/mL (range)
Subset 1
Subset 2
Subset 3
Combined
6 4 3861.5 (2962–7345) 36.5 (11–61) 154 84 70 Intensive No 7.07 (5.76–9.57)
8 7 4645 (3645–5443) 43 (23–62) 70 35 35 Sparse Yes 10.07 (0.95–16.28)
23 17 3444 (849–5171) 26 (3–46) 191 114 77 Intensive/Sparse Yes 4.11 (0.20–27.90)
37 28 3636 (849–7345) 31 (3–62) 415 233 182 – – 6.07 (0.20–27.90)
© 2014 John Wiley & Sons Ltd
Rifampin population pharmacokinetics in elephants 3
Simulations ModelRiskâ (Vose Software, Ghent, Belgium) was used to simulate dosage regimens based upon the pharmacokinetic parameters and BSV derived from the population pharmacokinetic models. Seven different doses were evaluated: 5, 10, 15, 20, 25, 30, and 35 mg/kg. A previously published AUC0–24/MIC ≥ 271 (fAUC0–24/MIC ≥ 41 based upon a median 15% free drug fraction) was used (Gumbo et al., 2007; Goutelle et al., 2009). For each scenario, 10 000 simulations were performed. All pharmacokinetic and pharmacodynamic parameters were considered normally or lognormal distributed for the simulations. Additional simulations using 5% and 10% free drug fraction were evaluated, based on a recent abstract (Lovering, 2013). A RSE of 5% was used for protein binding, consistent with the aforementioned studies. MICs simulated included the following: 0.01, 0.025, 0.05, 0.1, 0.5,1, and 2 lg/mL. A RSE of 30% was 10
8 7 6 5 4 3 2 1 0
0
5
10
15
20
25
Time
Data emp. prctile C.I 90%
16
1
Vi WTi ¼ Vstd F 3722 kg
Data emp. prctile prctile out. C.I 90% C.I 50% C.I 10% C.I out.
9
y1
including one- and two-compartment models with linear elimination and either zero- or first-order absorption, or a combination of zero- and first-order absorption. Structural models with and without a lag time and with transit compartments were examined. Multiple error models, including proportional, exponential, constant, and combined error models, were tested to describe residual unexplained variability (e). Relative standard error (RSE) values and goodness-of-fit plots were used in model discrimination. As rifampin was not administered parenterally, volume (V/F) and clearance (CL/F) are presented divided by F (oral bioavailability). Also, covariance between model parameters was evaluated. Model selection was based, in part, on a difference of more than 3.84 points in the objective function (OFV). Volume of distribution and CL estimates from a noncompartmental analysis (NCA) of Subset 1 provided initial parameter estimates for Vd and CL: 5350 L and 246 L/h. Tlag and ka were estimated. Initially, no covariates were included in the model. Covariates evaluated included the following: age, food, sex, and weight. Covariates found to be significant in univariate analysis were tested further by stepwise addition. Covariates were retained in the model if they had a change in OFV ≥ 3.84. Three population pharmacokinetic models were constructed. Model 1 consisted of only Subset 1. Model 2 consisted of Subset 1 and Subset 2. Model 3 combined all three subsets (Fig. 1). For all three models, allometric scaling was evaluated, with scaling to median body weight. For Vd, the exponent was tested fixed at 1 or estimated, and for CL, it was tested fixed at 0.75 or estimated (Wang et al., 2012). The following equations describe volume of distribution and clearance (3722 kg is used as an example):
C.I 50% C.I 10% C.I out.
14
;
12 10
CLi WTi ¼ Clstd F 3722 kg
0:75
y1
;
8 6 4
where Vstd and Clstd represent the standard V/F and CL/F estimates for the median subject, and WTi represents the weight of individual i. CLi and Vi represent the parameter estimates for individual i. A lognormal distribution was assumed for all population parameters.
2 0 0
5
10
15
20
Data emp. prctile prctile out C.I 90% C.I 50% C.I 10% C.I out.
25
20
Model 2
Model 3 y1
Model 1
Subset 1 Subset 1
25
Time
15
10
5
Subset 2
Subset 1
0
Subset 2 Fig. 1. Modeling overview. © 2014 John Wiley & Sons Ltd
Subset 3
0
5
10
15
20
25
Time
Fig. 2. Visual predictive check for oral rifampin concentrations (Models 1, 2, and 3, respectively).
4 E. F. Egelund et al.
Fig. 3. Diagnostic plots of Models 1, 2, and 3, respectively: population weighted residuals (PWRES), individual weighted residuals (IWRES), normalized prediction distribution error (NPDE) as a function of time and population prediction, probability density function (PDF), and quantile–quantile plots. © 2014 John Wiley & Sons Ltd
Rifampin population pharmacokinetics in elephants 5 Table 2. Population pharmacokinetic parameters of rifampin in 6 (Subset 1), 14 (Subsets 1 and 2), and 37 (all three subsets) elephants Model 1
PK parameter Tlag (h) ka (1/h) Vd (L) Cl (L/h)
Subset 1 Population Median (%SE) 0.711 0.737 4730 252
(33) (17) (21) (12)
Model 2
Subset 1 BSV (%SE) 0.789 0.322 0.172 0.219
Subset 1, 2 Population Median (%SE)
(29) (50) (37) (42)
0.345 0.504 6190 234
Model 3
Subsets 1, 2 BSV (%SE)
(34) (19) (9) (7)
0.924 0.377 0.085 0.080
(27) (38) (154) (136)
Subsets 1, 2, and 3 Population Median (%SE)
Subsets 1, 2, and 3 BSV (%SE)
0.496 (35) 0.526 (52) 3910 (33) 309 (25)
1.02 1.19 0.714 1.02
(28) (37) (31) (22)
BSV, between-subject variability.
added to account for uncertainty in the determination of the MICs.
RESULTS A one-compartment body model with first-order absorption with a lag time and first-order elimination best described rifampin pharmacokinetics. Depending on the initial parameter estimates, ‘flip-flop’ was seen between outputs of ke and ka. Clearance parameter estimates could be up to 10-fold higher (and subsequent half-life 10-fold shorter) depending on the initial estimates used. Similar discrepancies were found with the volume of distribution. The output using the NCA’s initial estimates is reported here. The combined error model best described the residual variability in Models 1 and 2, while the proportional error model best described the residual variability in Model 3. Individual weighted residuals (IWRES) plots showed predictions centered on zero with a constant variance. The inclusion of weight as a covariate for Vd was significant for Models 1 and 2. Estimating the exponent provided a better fit. Weight (log-transformed and centered on the median) provided the best size descriptor for Vd for Model 1 and Model 2, explaining 24% and 18% of the variability, respectively. For Model 3, weight did not have a significant effect on Vd and the
exponent was fixed to 1. Also for Model 3, food explained 30% of variability in Tlag. For model discrimination, visual predictive checks (Fig. 2), observed vs. predicted concentrations, as well as population and individual weighted residuals, and quantile–quantile plots (Fig. 3) showed an adequate fit for all three models. Good precision was determined for all of the parameter estimates. The RSE values were ≤35% for all estimates in Models 1 and 2, and 3 with the exception of ka for Model 3. In contrast, BSV was high across parameters. For Model 2, the BSV RSE was >100% for both Vd and CL. For Model 3, the BSV RSE was >100% for Tlag, ka, and CL (Table 2). Because of their greater precision, Model 1 parameter estimates were used in the simulations to determine the probability of target attainment (PTA). Figure 4 shows the PTA of AUC0–24/MIC ≥ 271 adjusted for 85%, 90%, and 95% protein binding. With 85% protein binding, a 10 mg/kg dose achieved 90% PTA up to an MIC of 0.25 lg/mL. With 90% and 95% protein binding, 10 mg/kg achieved 90% PTA at approximate MICs of 0.10 and 0.05 lg/mL, respectively.
DISCUSSION We describe the population pharmacokinetics of oral rifampin using data for 37 elephants from three separate data sets. RIF 10 mg/kg
100 90 80
Target attainement %
70 60
Protein.binding 85% protein binding 90% protein binding 95% protein binding
50 40 30
Fig. 4. Probabilities of target attainment (PTA): percentage of elephants achieving an AUC/MIC0–24 > 271 at 10 mg/kg, using free rifampin concentrations in serum at 85%, 90%, and 95% protein binding. © 2014 John Wiley & Sons Ltd
20 10 0 0.01
0.025
0.05
0.1 0.25 MIC (µg/mL)
0.5
1
2
6 E. F. Egelund et al.
‘Flip-flop’ was encountered, but managed with careful selection of initial estimates. Subset 1 consisted of single-dose data. In humans, rifampin undergoes auto-induction, and half-life declines from 3–4 h in single-dose studies to 1–2 h under steady-state conditions (Acocella et al., 1971; Loos et al., 1985). It is not known if rifampin auto-induction occurs in elephants, but this could not be assessed with Subset 1 data. Only sparse data were available at steady-state, and most of that data were collected close to the time of dosing, thus prohibiting an assessment of auto-induction. The large variability in ka made it difficult to model the absorption process. Rifampin absorption is known to be variable in humans (van Crevel et al., 2002; McIlleron et al., 2006; Wilkins et al., 2008). Although the medications were not administered with food, food was available ad libitum to elephants in Subset 1, and thus, the amount, timing, and effect of food intake on plasma concentrations could not be assessed accurately. In Model 3, food was a significant covariate for the lag time between elephants that ingested food with rifampin (lag time = 2.69 h) vs. those that did not (lag time = 0.496 h). Food explained 30% of BSV. To maximize rifampin’s absorption, it should not be administered with food (Peloquin et al., 1999; Wilkins et al., 2008). Model 1 estimates were chosen for simulations because they showed greater precision (see Table 2). Recent studies indicate AUC/MIC is the PK/PD parameter most associated with antimicrobial killing (Gumbo et al., 2007). Simulations using an AUC0–24/MIC of 271 (subsequently adjusted for protein binding) suggest the current dose of 10 mg/kg is adequate for MICs up to 0.25 lg/mL (MIC range 0.15–0.50 lg/mL), and these are typical clinical MIC values (Lee & Heifets, 1987). The wide variability in the pharmacokinetic parameters underscores the need for close monitoring of drug concentrations (i.e., TDM) following initial dosing. The median Cmax for the combined subsets was 6.07 lg/mL, which is below the targeted 8–24 lg/mL range. Subsets 2 and 3 consisted of TDM data which may not have captured the Cmax for multiple reasons: (i) a limited sampling strategy, (ii) data from the intensive data set (Subset 1) indicate Cmax is best captured between 6 and 8 h rather than 2 to 3 h, and (iii) multiple oral formulations were used which affects bioavailability (Brock et al., 2014). The median Cmax of 7.07 from Subset 1 more accurately reflects the Cmax experienced with current dosing but also is below the targeted range. Our study has a number of limitations. BLQ values may be less precise and may introduce bias into the model. Next, we could not quantify the food effect on rifampin pharmacokinetics. Third, interoccasion variability could not be estimated from these data (Bonate, 2010). Finally, the assumption of allometric scaling, while commonly used in population pharmacokinetic modeling, may not always reflect physiological factors.
CONCLUSION Rifampin plasma concentrations were highly variable between elephants. Some of this variability is caused by food consumption.
We recommend dosing rifampin without food. Weight-based dosing is warranted. The median Cmax in elephants was below the usual range for humans. However, given the longer RIF half-life in elephants and targeting AUC/MIC, the current recommended dose of 10 mg/kg appears to be adequate, provided that the isolate has an MIC of ≤0.25 lg/mL. Further studies are needed to determine the protein binding of rifampin in elephants. Finally, outcome data are needed to characterize the association between the attainment of selected AUC/MIC values and cure.
ACKNOWLEDGMENTS The authors thank the chemists at the Infectious Disease Pharmacokinetics Laboratory at the University of Florida for sample analysis: Kyung-Mee Kim, Behrang Mahjoub, Vaneska Mayor, and Theodore Zagurski. We also thank Dr. Aline Barth for assistance with simulations.
REFERENCES Acocella, G. (1978) Clinical pharmacokinetics of rifampicin. Clinical Pharmacokinetics, 3, 108–127. Acocella, G., Pagani, V., Marchetti, M., Baroni, G.C. & Nicolis, F.B. (1971) Kinetic studies on rifampicin. I. Serum concentration analysis in subjects treated with different oral doses over a period of two weeks. Chemotherapy, 16, 356–370. Ahn, J.E., Karlsson, M.O., Dunne, A. & Ludden, T.M. (2008) Likelihood based approaches to handling data below the quantification limit using NONMEM VI. Journal of Pharmacokinetics and Pharmacodynamics, 35, 401–421. Bonate, P.L. (2010) Pharmacokinetic-Pharmacodynamic Modeling and Simulation. Springer Science, New York, NY. Brock, A.P., Isaza, R., Egelund, E.F., Hunter, R.P. & Peloquin, C.A. (2014) The pharmacokinetics of a single oral or rectal dose of concurrently administered isoniazid, rifampin, pyrazinamide, and ethambutol in Asian elephants (Elephas maximus). Journal of Veterinary Pharmacology and Therapeutics, 37, 472–479. van Crevel, R., Alisjahbana, B., de Lange, W.C., Borst, F., Danusantoso, H., van der Meer, J.W., Burger, D. & Nelwan, R.H. (2002) Low plasma concentrations of rifampicin in tuberculosis patients in Indonesia. The International Journal of Tuberculosis and Lung Disease, 6, 497–502. Dumonceaux, G.A., St Leger, J., Olsen, J.H., Burton, M.S., Ashkin, D. & Maslow, J.N. (2011) Genitourinary and pulmonary multidrug resistant Mycobacterium tuberculosis infection in an Asian elephant (Elephas maximus). Journal of Zoo and Wildlife Medicine, 42, 709–712. Duval, V. & Karlsson, M.O. (2002) Impact of omission or replacement of data below the limit of quantification on parameter estimates in a two-compartment model. Pharmaceutical Research, 19, 1835–1840. Goutelle, S., Bourguignon, L., Maire, P.H., Van Guilder, M., Conte, J.E. Jr & Jelliffe, R.W. (2009) Population modeling and Monte Carlo simulation study of the pharmacokinetics and antituberculosis pharmacodynamics of rifampin in lungs. Antimicrobial Agents and Chemotherapy, 53, 2974–2981. Gumbo, T., Louie, A., Deziel, M.R., Liu, W., Parsons, L.M., Salfinger, M. & Drusano, G.L. (2007) Concentration-dependent Mycobacterium tuberculosis killing and prevention of resistance by rifampin. Antimicrobial Agents and Chemotherapy, 51, 3781–3788. Jayaram, R., Gaonkar, S., Kaur, P., Suresh, B.L., Mahesh, B.N., Jayashree, R., Nandi, V., Bharat, S., Shandil, R.K., Kantharaj, E. & Bala-
© 2014 John Wiley & Sons Ltd
Rifampin population pharmacokinetics in elephants 7 subramanian, V. (2003) Pharmacokinetics-pharmacodynamics of rifampin in an aerosol infection model of tuberculosis. Antimicrobial Agents and Chemotherapy, 47, 2118–2124. Lee, C.N. & Heifets, L.B. (1987) Determination of minimal inhibitory concentrations of antituberculosis drugs by radiometric and conventional methods. The American Review of Respiratory Disease, 136, 349 –352. Loos, U., Musch, E., Jensen, J.C., Mikus, G., Schwabe, H.K. & Eichelbaum, M. (1985) Pharmacokinetics of oral and intravenous rifampicin during chronic administration. Klinische Wochenschrift, 63, 1205–1211. Lovering, A.M.H.J. (2013) Rifampicin Protein Binding in Patients with Mycobacterial Infection; a Comparison of Ultrafiltration and Rapid Equilibrium Dialysis. 53rd Interscience Conference on Antimicrobial Agents and Chemotherapy, Denver, CO. Maslow, J.N., Mikota, S.K., Zhu, M., Isaza, R., Peddie, L.R., Dunker, F., Peddie, J., Riddle, H. & Peloquin, C.A. (2005a) Population pharmacokinetics of isoniazid in the treatment of Mycobacterium tuberculosis among Asian and African elephants (Elephas maximus and Loxodonta africana). Journal of Veterinary Pharmacology and Therapeutics, 28, 21–27. Maslow, J.N., Mikota, S.K., Zhu, M., Riddle, H. & Peloquin, C.A. (2005b) Pharmacokinetics of ethambutol (EMB) in elephants. Journal of Veterinary Pharmacology and Therapeutics, 28, 321–323. McGaughey, C. (1961) Diseases of elephants, Part III. Ceylon Veterinary Journal, 9, 94–98. McIlleron, H., Wash, P., Burger, A., Norman, J., Folb, P.I. & Smith, P. (2006) Determinants of rifampin, isoniazid, pyrazinamide, and ethambutol pharmacokinetics in a cohort of tuberculosis patients. Antimicrobial Agents and Chemotherapy, 50, 1170–1177. Nuermberger, E. & Grosset, J. (2004) Pharmacokinetic and pharmacodynamic issues in the treatment of mycobacterial infections. European Journal of Clinical Microbiology and Infectious Diseases, 23, 243–255.
© 2014 John Wiley & Sons Ltd
Peloquin, C.A., Namdar, R., Singleton, M.D. & Nix, D.E. (1999) Pharmacokinetics of rifampin under fasting conditions, with food, and with antacids. Chest, 115, 12–18. Peloquin, C.A., Maslow, J.N., Mikota, S.K., Forrest, A., Dunker, F., Isaza, R., Peddie, L.R., Peddie, J. & Zhu, M. (2006) Dose selection and pharmacokinetics of rifampin in elephants for the treatment of tuberculosis. Journal of Veterinary Pharmacology and Therapeutics, 29, 581–585. USDA. (2008) Guidelines for the Control of Tuberculosis in Elephants 2008. United States Department of Agriculture, Animal and Plant Health Inspection Service, Animal Care Policies, From the National Tuberculosis Working Group for Zoo and Wildlife Species Washington D.C. http://www.aphis.usda.gov/animal_welfare/ downloads/elephant/elephant_tb.pdf. Verbist, L. (1969) Rifampicin activity “in vitro” and in established tuberculosis in mice. Acta Tuberculosea et Pneumologica Belgica, 60, 397–412. Wang, C., Peeters, M.Y., Allegaert, K., Blusse van Oud-Alblas, H.J., Krekels, E.H., Tibboel, D., Danhof, M. & Knibbe, C.A. (2012) A bodyweight-dependent allometric exponent for scaling clearance across the human life-span. Pharmaceutical Research, 29, 1570– 1581. Wilkins, J.J., Savic, R.M., Karlsson, M.O., Langdon, G., McIlleron, H., Pillai, G., Smith, P.J. & Simonsson, U.S. (2008) Population pharmacokinetics of rifampin in pulmonary tuberculosis patients, including a semimechanistic model to describe variable absorption. Antimicrobial Agents and Chemotherapy, 52, 2138–2148. Zhu, M., Maslow, J.N., Mikota, S.K., Isaza, R., Dunker, F., Riddle, H. & Peloquin, C.A. (2005) Population pharmacokinetics of pyrazinamide in elephants. Journal of Veterinary Pharmacology and Therapeutics, 28, 403–409.
Population pharmacokinetics of rifampin in the treatment of Mycobacterium tuberculosis in Asian elephants E. F. EGELUND* R. ISAZA
†
A. P. BROCK †
Egelund, E. F., Isaza, R., Brock, A. P., Alsultan, A., An, G., Peloquin, C. A. Population pharmacokinetics of rifampin in the treatment of Mycobacterium tuberculosis in Asian elephants. J. vet. Pharmacol. Therap. doi: 10.1111/jvp. 12156.
A. ALSULTAN* G. AN ‡ & C. A. PELOQUIN* *Department of Pharmacotherapy and Translational Research, College of Pharmacy, University of Florida, Gainesville, FL, USA; † Department of Small Animal Clinical Sciences, College of Veterinary Medicine, University of Florida, Gainesville, FL, USA; ‡ Department of Pharmaceutics, College of Pharmacy, University of Florida, Orlando, FL, USA
The objective of this study was to develop a population pharmacokinetic model for rifampin in elephants. Rifampin concentration data from three sources were pooled to provide a total of 233 oral concentrations from 37 Asian elephants. The population pharmacokinetic models were created using Monolix (version 4.2). Simulations were conducted using ModelRisk. We examined the influence of age, food, sex, and weight as model covariates. We further optimized the dosing of rifampin based upon simulations using the population pharmacokinetic model. Rifampin pharmacokinetics were best described by a one-compartment open model including first-order absorption with a lag time and first-order elimination. Body weight was a significant covariate for volume of distribution, and food intake was a significant covariate for lag time. The median Cmax of 6.07 lg/mL was below the target range of 8–24 lg/mL. Monte Carlo simulations predicted the highest treatable MIC of 0.25 lg/mL with the current initial dosing recommendation of 10 mg/kg, based upon a previously published target AUC0–24/MIC > 271 (fAUC > 41). Simulations from the population model indicate that the current dose of 10 mg/kg may be adequate for MICs up to 0.25 lg/mL. While the targeted AUC/MIC may be adequate for most MICs, the median Cmax for all elephants is below the human and elephant targeted ranges. (Paper received 21 May 2014; accepted for publication 10 July 2014) Charles Peloquin, PharmD, FCCP, Professor, and Director, Infectious Disease Pharmacokinetics Laboratory, College of Pharmacy, and Emerging Pathogens Institute, University of Florida, 1600 SW Archer Road, Rm P4-33, PO Box 100486, Gainesville, FL 32610-0486, USA. E-mail: [email protected]
INTRODUCTION Mycobacterium tuberculosis (MTb), the causative agent of tuberculosis (TB), is believed to have affected elephant populations for centuries (McGaughey, 1961). Limited data exist regarding the appropriate treatment of MTb in elephants. Currently, elephants are treated with the same drug regimen used in humans, with doses based on linear scaling of weight (USDA, 2008). Only recently have pharmacokinetic studies been conducted to determine the concentration–time profiles and pharmacokinetic parameters associated with the doses used in the treatment of TB in elephants (Maslow et al., 2005a,b; Zhu et al., 2005; Peloquin et al., 2006; Brock et al., 2014). At the time of this writing, only one previous paper exists which examines the population pharmacokinetics of rifampin in elephants (Peloquin et al., 2006). © 2014 John Wiley & Sons Ltd
Along with isoniazid, rifampin is the most important drug for treating tuberculosis, in both humans and elephants. However, the plasma concentrations required for efficacy in elephants remain unknown. Currently, dose recommendations are based on human plasma concentrations, with a target maximum concentration (Cmax) range of 8–24 lg/mL (Acocella, 1978; USDA, 2008). This target Cmax range lies well above the minimum inhibitory concentration (MIC) range typically seen with MTb (0.15–0.5 lg/mL). Ideally, the initial recommended rifampin dose for elephants is 10 mg/kg, to be followed by therapeutic drug monitoring (TDM) (USDA, 2008). Only sparse concentration data were available previously, making it difficult to assess the current recommended dose in relation to the drug’s Cmax. Several published studies confirm that rifampin’s antibacterial effect is concentration dependent, requiring an optimization of either Cmax or the area under the 1
2 E. F. Egelund et al.
curve (AUC), that is, the drug’s exposure, over the drug’s MIC to maximize the killing of MTb (Verbist, 1969; Jayaram et al., 2003; Nuermberger & Grosset, 2004; Gumbo et al., 2007). Recent publications have cited the AUC/MIC as the pharmacokinetic/pharmacodynamic (PK/PD) parameter most highly correlated with killing, while Cmax/MIC is most associated with prevention of resistance and postantibiotic effect (Jayaram et al., 2003; Gumbo et al., 2007). Additionally, covariates that could influence the between-elephant PK variability have not been identified. A greater understanding of these factors is needed to optimize treatment regimens and prevent resistance (Dumonceaux et al., 2011). Nonlinear mixed effect (NLME) models enable investigators to determine between-subject variability (BSV), accounting for both fixed and random effects. Fixed effects are population parameters which are considered constant (e.g., Vd, Cl) while random effects refer to unexplained variation. NLME allows the utilization of both intensive and sparse data. The aim of this study was to develop a population pharmacokinetic model using available intensive and sparse data to examine the influence of specific covariates on rifampin’s pharmacokinetics. Further, we assessed the current dosing recommendations using simulations based upon our population model.
although some had more than two concentration–time points. Dosing at 10 mg/kg was either rectal or oral. Subset 3 consisted of 214 concentrations from 23 elephants. Elephants were both intensively and sparsely sampled after either rectal (two methods) or oral administration (four methods) with or without food, as previously described (Peloquin et al., 2006). Dosing ranged from 4 to 12 mg/kg. Sample collection, storage, and analysis For each subset, samples were collected from the caudal auricular vein into lithium heparin tubes. Samples then were placed on ice until centrifugation. Plasma was harvested and stored on dry ice prior to transfer for up to 24 h and then stored at 80 °C until assayed. Prior to analysis, samples were allowed to thaw at room temperature. For each subset, concentrations were determined by high-performance liquid chromatography using a validated assay described previously (Peloquin et al., 1999). Briefly, the rifampin standard curve ranged from 0.5 to 50 lg/mL. The lower limit of quantification (LLOQ) was the lowest point of the standard curve (i.e., 0.5 lg/mL). All assays were free of any interfering peaks from other study drugs and certified by the College of American Pathologists. Pharmacokinetic data analysis
MATERIALS AND METHODS Study population Data from three sources were pooled to form the data set used in this analysis. Four hundred and fifteen samples (233 oral, 182 rectal) were collected from 37 elephants, weighing 849– 7345 kg (see Table 1 for full demographics). Subset 1 consisted of 154 concentrations intensively sampled from six elephants. Plasma samples were collected at 14 time points. Rifampin was administered orally at 10 mg/kg to six elephants and rectally at 10 mg/kg to five of these elephants both as a single dose. Elephants were allowed food and water ad libitum (Brock et al., 2014). Subset 2 consisted of 62 concentrations from eight elephants on different occasions. Elephants were sparsely sampled, primarily at 2 and 4 h after drug administration for TDM,
A majority of plasma concentrations following rectal administration were below the limit of quantification (BLQ), thus all rectal concentrations were excluded. A number of concentrations in the oral data set also were BLQ. Potential data handling options for BLQ values include using the raw data, replacing the data with half the BLQ, omitting the BLQ data, or using likelihood-based approaches (Ahn et al., 2008). Each of these methods can introduce bias to the modeling process (Duval & Karlsson, 2002). We elected to include the BLQ data because most were within 10% of the LLOQ. Further, data from each subject were graphed and evaluated to verify that any BLQ data were pharmacokinetically plausible. A compartmental model using NLME modeling was developed to characterize pharmacokinetic parameters. The software, Monolix v 4.2 (Lixoft. Orsay, France), was used to analyze the data. Several structural models were evaluated,
Table 1. Summary of subject demographics (median and range) and rifampin sample data (median and range) for purposes of a population pharmacokinetic model in Asian elephants Characteristics No. of subjects Female sex Median weight in kg (range) Median age (range) No. of samples Oral Rectal Sampling Steady-state Oral median Cmax in lg/mL (range)
Subset 1
Subset 2
Subset 3
Combined
6 4 3861.5 (2962–7345) 36.5 (11–61) 154 84 70 Intensive No 7.07 (5.76–9.57)
8 7 4645 (3645–5443) 43 (23–62) 70 35 35 Sparse Yes 10.07 (0.95–16.28)
23 17 3444 (849–5171) 26 (3–46) 191 114 77 Intensive/Sparse Yes 4.11 (0.20–27.90)
37 28 3636 (849–7345) 31 (3–62) 415 233 182 – – 6.07 (0.20–27.90)
© 2014 John Wiley & Sons Ltd
Rifampin population pharmacokinetics in elephants 3
Simulations ModelRiskâ (Vose Software, Ghent, Belgium) was used to simulate dosage regimens based upon the pharmacokinetic parameters and BSV derived from the population pharmacokinetic models. Seven different doses were evaluated: 5, 10, 15, 20, 25, 30, and 35 mg/kg. A previously published AUC0–24/MIC ≥ 271 (fAUC0–24/MIC ≥ 41 based upon a median 15% free drug fraction) was used (Gumbo et al., 2007; Goutelle et al., 2009). For each scenario, 10 000 simulations were performed. All pharmacokinetic and pharmacodynamic parameters were considered normally or lognormal distributed for the simulations. Additional simulations using 5% and 10% free drug fraction were evaluated, based on a recent abstract (Lovering, 2013). A RSE of 5% was used for protein binding, consistent with the aforementioned studies. MICs simulated included the following: 0.01, 0.025, 0.05, 0.1, 0.5,1, and 2 lg/mL. A RSE of 30% was 10
8 7 6 5 4 3 2 1 0
0
5
10
15
20
25
Time
Data emp. prctile C.I 90%
16
1
Vi WTi ¼ Vstd F 3722 kg
Data emp. prctile prctile out. C.I 90% C.I 50% C.I 10% C.I out.
9
y1
including one- and two-compartment models with linear elimination and either zero- or first-order absorption, or a combination of zero- and first-order absorption. Structural models with and without a lag time and with transit compartments were examined. Multiple error models, including proportional, exponential, constant, and combined error models, were tested to describe residual unexplained variability (e). Relative standard error (RSE) values and goodness-of-fit plots were used in model discrimination. As rifampin was not administered parenterally, volume (V/F) and clearance (CL/F) are presented divided by F (oral bioavailability). Also, covariance between model parameters was evaluated. Model selection was based, in part, on a difference of more than 3.84 points in the objective function (OFV). Volume of distribution and CL estimates from a noncompartmental analysis (NCA) of Subset 1 provided initial parameter estimates for Vd and CL: 5350 L and 246 L/h. Tlag and ka were estimated. Initially, no covariates were included in the model. Covariates evaluated included the following: age, food, sex, and weight. Covariates found to be significant in univariate analysis were tested further by stepwise addition. Covariates were retained in the model if they had a change in OFV ≥ 3.84. Three population pharmacokinetic models were constructed. Model 1 consisted of only Subset 1. Model 2 consisted of Subset 1 and Subset 2. Model 3 combined all three subsets (Fig. 1). For all three models, allometric scaling was evaluated, with scaling to median body weight. For Vd, the exponent was tested fixed at 1 or estimated, and for CL, it was tested fixed at 0.75 or estimated (Wang et al., 2012). The following equations describe volume of distribution and clearance (3722 kg is used as an example):
C.I 50% C.I 10% C.I out.
14
;
12 10
CLi WTi ¼ Clstd F 3722 kg
0:75
y1
;
8 6 4
where Vstd and Clstd represent the standard V/F and CL/F estimates for the median subject, and WTi represents the weight of individual i. CLi and Vi represent the parameter estimates for individual i. A lognormal distribution was assumed for all population parameters.
2 0 0
5
10
15
20
Data emp. prctile prctile out C.I 90% C.I 50% C.I 10% C.I out.
25
20
Model 2
Model 3 y1
Model 1
Subset 1 Subset 1
25
Time
15
10
5
Subset 2
Subset 1
0
Subset 2 Fig. 1. Modeling overview. © 2014 John Wiley & Sons Ltd
Subset 3
0
5
10
15
20
25
Time
Fig. 2. Visual predictive check for oral rifampin concentrations (Models 1, 2, and 3, respectively).
4 E. F. Egelund et al.
Fig. 3. Diagnostic plots of Models 1, 2, and 3, respectively: population weighted residuals (PWRES), individual weighted residuals (IWRES), normalized prediction distribution error (NPDE) as a function of time and population prediction, probability density function (PDF), and quantile–quantile plots. © 2014 John Wiley & Sons Ltd
Rifampin population pharmacokinetics in elephants 5 Table 2. Population pharmacokinetic parameters of rifampin in 6 (Subset 1), 14 (Subsets 1 and 2), and 37 (all three subsets) elephants Model 1
PK parameter Tlag (h) ka (1/h) Vd (L) Cl (L/h)
Subset 1 Population Median (%SE) 0.711 0.737 4730 252
(33) (17) (21) (12)
Model 2
Subset 1 BSV (%SE) 0.789 0.322 0.172 0.219
Subset 1, 2 Population Median (%SE)
(29) (50) (37) (42)
0.345 0.504 6190 234
Model 3
Subsets 1, 2 BSV (%SE)
(34) (19) (9) (7)
0.924 0.377 0.085 0.080
(27) (38) (154) (136)
Subsets 1, 2, and 3 Population Median (%SE)
Subsets 1, 2, and 3 BSV (%SE)
0.496 (35) 0.526 (52) 3910 (33) 309 (25)
1.02 1.19 0.714 1.02
(28) (37) (31) (22)
BSV, between-subject variability.
added to account for uncertainty in the determination of the MICs.
RESULTS A one-compartment body model with first-order absorption with a lag time and first-order elimination best described rifampin pharmacokinetics. Depending on the initial parameter estimates, ‘flip-flop’ was seen between outputs of ke and ka. Clearance parameter estimates could be up to 10-fold higher (and subsequent half-life 10-fold shorter) depending on the initial estimates used. Similar discrepancies were found with the volume of distribution. The output using the NCA’s initial estimates is reported here. The combined error model best described the residual variability in Models 1 and 2, while the proportional error model best described the residual variability in Model 3. Individual weighted residuals (IWRES) plots showed predictions centered on zero with a constant variance. The inclusion of weight as a covariate for Vd was significant for Models 1 and 2. Estimating the exponent provided a better fit. Weight (log-transformed and centered on the median) provided the best size descriptor for Vd for Model 1 and Model 2, explaining 24% and 18% of the variability, respectively. For Model 3, weight did not have a significant effect on Vd and the
exponent was fixed to 1. Also for Model 3, food explained 30% of variability in Tlag. For model discrimination, visual predictive checks (Fig. 2), observed vs. predicted concentrations, as well as population and individual weighted residuals, and quantile–quantile plots (Fig. 3) showed an adequate fit for all three models. Good precision was determined for all of the parameter estimates. The RSE values were ≤35% for all estimates in Models 1 and 2, and 3 with the exception of ka for Model 3. In contrast, BSV was high across parameters. For Model 2, the BSV RSE was >100% for both Vd and CL. For Model 3, the BSV RSE was >100% for Tlag, ka, and CL (Table 2). Because of their greater precision, Model 1 parameter estimates were used in the simulations to determine the probability of target attainment (PTA). Figure 4 shows the PTA of AUC0–24/MIC ≥ 271 adjusted for 85%, 90%, and 95% protein binding. With 85% protein binding, a 10 mg/kg dose achieved 90% PTA up to an MIC of 0.25 lg/mL. With 90% and 95% protein binding, 10 mg/kg achieved 90% PTA at approximate MICs of 0.10 and 0.05 lg/mL, respectively.
DISCUSSION We describe the population pharmacokinetics of oral rifampin using data for 37 elephants from three separate data sets. RIF 10 mg/kg
100 90 80
Target attainement %
70 60
Protein.binding 85% protein binding 90% protein binding 95% protein binding
50 40 30
Fig. 4. Probabilities of target attainment (PTA): percentage of elephants achieving an AUC/MIC0–24 > 271 at 10 mg/kg, using free rifampin concentrations in serum at 85%, 90%, and 95% protein binding. © 2014 John Wiley & Sons Ltd
20 10 0 0.01
0.025
0.05
0.1 0.25 MIC (µg/mL)
0.5
1
2
6 E. F. Egelund et al.
‘Flip-flop’ was encountered, but managed with careful selection of initial estimates. Subset 1 consisted of single-dose data. In humans, rifampin undergoes auto-induction, and half-life declines from 3–4 h in single-dose studies to 1–2 h under steady-state conditions (Acocella et al., 1971; Loos et al., 1985). It is not known if rifampin auto-induction occurs in elephants, but this could not be assessed with Subset 1 data. Only sparse data were available at steady-state, and most of that data were collected close to the time of dosing, thus prohibiting an assessment of auto-induction. The large variability in ka made it difficult to model the absorption process. Rifampin absorption is known to be variable in humans (van Crevel et al., 2002; McIlleron et al., 2006; Wilkins et al., 2008). Although the medications were not administered with food, food was available ad libitum to elephants in Subset 1, and thus, the amount, timing, and effect of food intake on plasma concentrations could not be assessed accurately. In Model 3, food was a significant covariate for the lag time between elephants that ingested food with rifampin (lag time = 2.69 h) vs. those that did not (lag time = 0.496 h). Food explained 30% of BSV. To maximize rifampin’s absorption, it should not be administered with food (Peloquin et al., 1999; Wilkins et al., 2008). Model 1 estimates were chosen for simulations because they showed greater precision (see Table 2). Recent studies indicate AUC/MIC is the PK/PD parameter most associated with antimicrobial killing (Gumbo et al., 2007). Simulations using an AUC0–24/MIC of 271 (subsequently adjusted for protein binding) suggest the current dose of 10 mg/kg is adequate for MICs up to 0.25 lg/mL (MIC range 0.15–0.50 lg/mL), and these are typical clinical MIC values (Lee & Heifets, 1987). The wide variability in the pharmacokinetic parameters underscores the need for close monitoring of drug concentrations (i.e., TDM) following initial dosing. The median Cmax for the combined subsets was 6.07 lg/mL, which is below the targeted 8–24 lg/mL range. Subsets 2 and 3 consisted of TDM data which may not have captured the Cmax for multiple reasons: (i) a limited sampling strategy, (ii) data from the intensive data set (Subset 1) indicate Cmax is best captured between 6 and 8 h rather than 2 to 3 h, and (iii) multiple oral formulations were used which affects bioavailability (Brock et al., 2014). The median Cmax of 7.07 from Subset 1 more accurately reflects the Cmax experienced with current dosing but also is below the targeted range. Our study has a number of limitations. BLQ values may be less precise and may introduce bias into the model. Next, we could not quantify the food effect on rifampin pharmacokinetics. Third, interoccasion variability could not be estimated from these data (Bonate, 2010). Finally, the assumption of allometric scaling, while commonly used in population pharmacokinetic modeling, may not always reflect physiological factors.
CONCLUSION Rifampin plasma concentrations were highly variable between elephants. Some of this variability is caused by food consumption.
We recommend dosing rifampin without food. Weight-based dosing is warranted. The median Cmax in elephants was below the usual range for humans. However, given the longer RIF half-life in elephants and targeting AUC/MIC, the current recommended dose of 10 mg/kg appears to be adequate, provided that the isolate has an MIC of ≤0.25 lg/mL. Further studies are needed to determine the protein binding of rifampin in elephants. Finally, outcome data are needed to characterize the association between the attainment of selected AUC/MIC values and cure.
ACKNOWLEDGMENTS The authors thank the chemists at the Infectious Disease Pharmacokinetics Laboratory at the University of Florida for sample analysis: Kyung-Mee Kim, Behrang Mahjoub, Vaneska Mayor, and Theodore Zagurski. We also thank Dr. Aline Barth for assistance with simulations.
REFERENCES Acocella, G. (1978) Clinical pharmacokinetics of rifampicin. Clinical Pharmacokinetics, 3, 108–127. Acocella, G., Pagani, V., Marchetti, M., Baroni, G.C. & Nicolis, F.B. (1971) Kinetic studies on rifampicin. I. Serum concentration analysis in subjects treated with different oral doses over a period of two weeks. Chemotherapy, 16, 356–370. Ahn, J.E., Karlsson, M.O., Dunne, A. & Ludden, T.M. (2008) Likelihood based approaches to handling data below the quantification limit using NONMEM VI. Journal of Pharmacokinetics and Pharmacodynamics, 35, 401–421. Bonate, P.L. (2010) Pharmacokinetic-Pharmacodynamic Modeling and Simulation. Springer Science, New York, NY. Brock, A.P., Isaza, R., Egelund, E.F., Hunter, R.P. & Peloquin, C.A. (2014) The pharmacokinetics of a single oral or rectal dose of concurrently administered isoniazid, rifampin, pyrazinamide, and ethambutol in Asian elephants (Elephas maximus). Journal of Veterinary Pharmacology and Therapeutics, 37, 472–479. van Crevel, R., Alisjahbana, B., de Lange, W.C., Borst, F., Danusantoso, H., van der Meer, J.W., Burger, D. & Nelwan, R.H. (2002) Low plasma concentrations of rifampicin in tuberculosis patients in Indonesia. The International Journal of Tuberculosis and Lung Disease, 6, 497–502. Dumonceaux, G.A., St Leger, J., Olsen, J.H., Burton, M.S., Ashkin, D. & Maslow, J.N. (2011) Genitourinary and pulmonary multidrug resistant Mycobacterium tuberculosis infection in an Asian elephant (Elephas maximus). Journal of Zoo and Wildlife Medicine, 42, 709–712. Duval, V. & Karlsson, M.O. (2002) Impact of omission or replacement of data below the limit of quantification on parameter estimates in a two-compartment model. Pharmaceutical Research, 19, 1835–1840. Goutelle, S., Bourguignon, L., Maire, P.H., Van Guilder, M., Conte, J.E. Jr & Jelliffe, R.W. (2009) Population modeling and Monte Carlo simulation study of the pharmacokinetics and antituberculosis pharmacodynamics of rifampin in lungs. Antimicrobial Agents and Chemotherapy, 53, 2974–2981. Gumbo, T., Louie, A., Deziel, M.R., Liu, W., Parsons, L.M., Salfinger, M. & Drusano, G.L. (2007) Concentration-dependent Mycobacterium tuberculosis killing and prevention of resistance by rifampin. Antimicrobial Agents and Chemotherapy, 51, 3781–3788. Jayaram, R., Gaonkar, S., Kaur, P., Suresh, B.L., Mahesh, B.N., Jayashree, R., Nandi, V., Bharat, S., Shandil, R.K., Kantharaj, E. & Bala-
© 2014 John Wiley & Sons Ltd
Rifampin population pharmacokinetics in elephants 7 subramanian, V. (2003) Pharmacokinetics-pharmacodynamics of rifampin in an aerosol infection model of tuberculosis. Antimicrobial Agents and Chemotherapy, 47, 2118–2124. Lee, C.N. & Heifets, L.B. (1987) Determination of minimal inhibitory concentrations of antituberculosis drugs by radiometric and conventional methods. The American Review of Respiratory Disease, 136, 349 –352. Loos, U., Musch, E., Jensen, J.C., Mikus, G., Schwabe, H.K. & Eichelbaum, M. (1985) Pharmacokinetics of oral and intravenous rifampicin during chronic administration. Klinische Wochenschrift, 63, 1205–1211. Lovering, A.M.H.J. (2013) Rifampicin Protein Binding in Patients with Mycobacterial Infection; a Comparison of Ultrafiltration and Rapid Equilibrium Dialysis. 53rd Interscience Conference on Antimicrobial Agents and Chemotherapy, Denver, CO. Maslow, J.N., Mikota, S.K., Zhu, M., Isaza, R., Peddie, L.R., Dunker, F., Peddie, J., Riddle, H. & Peloquin, C.A. (2005a) Population pharmacokinetics of isoniazid in the treatment of Mycobacterium tuberculosis among Asian and African elephants (Elephas maximus and Loxodonta africana). Journal of Veterinary Pharmacology and Therapeutics, 28, 21–27. Maslow, J.N., Mikota, S.K., Zhu, M., Riddle, H. & Peloquin, C.A. (2005b) Pharmacokinetics of ethambutol (EMB) in elephants. Journal of Veterinary Pharmacology and Therapeutics, 28, 321–323. McGaughey, C. (1961) Diseases of elephants, Part III. Ceylon Veterinary Journal, 9, 94–98. McIlleron, H., Wash, P., Burger, A., Norman, J., Folb, P.I. & Smith, P. (2006) Determinants of rifampin, isoniazid, pyrazinamide, and ethambutol pharmacokinetics in a cohort of tuberculosis patients. Antimicrobial Agents and Chemotherapy, 50, 1170–1177. Nuermberger, E. & Grosset, J. (2004) Pharmacokinetic and pharmacodynamic issues in the treatment of mycobacterial infections. European Journal of Clinical Microbiology and Infectious Diseases, 23, 243–255.
© 2014 John Wiley & Sons Ltd
Peloquin, C.A., Namdar, R., Singleton, M.D. & Nix, D.E. (1999) Pharmacokinetics of rifampin under fasting conditions, with food, and with antacids. Chest, 115, 12–18. Peloquin, C.A., Maslow, J.N., Mikota, S.K., Forrest, A., Dunker, F., Isaza, R., Peddie, L.R., Peddie, J. & Zhu, M. (2006) Dose selection and pharmacokinetics of rifampin in elephants for the treatment of tuberculosis. Journal of Veterinary Pharmacology and Therapeutics, 29, 581–585. USDA. (2008) Guidelines for the Control of Tuberculosis in Elephants 2008. United States Department of Agriculture, Animal and Plant Health Inspection Service, Animal Care Policies, From the National Tuberculosis Working Group for Zoo and Wildlife Species Washington D.C. http://www.aphis.usda.gov/animal_welfare/ downloads/elephant/elephant_tb.pdf. Verbist, L. (1969) Rifampicin activity “in vitro” and in established tuberculosis in mice. Acta Tuberculosea et Pneumologica Belgica, 60, 397–412. Wang, C., Peeters, M.Y., Allegaert, K., Blusse van Oud-Alblas, H.J., Krekels, E.H., Tibboel, D., Danhof, M. & Knibbe, C.A. (2012) A bodyweight-dependent allometric exponent for scaling clearance across the human life-span. Pharmaceutical Research, 29, 1570– 1581. Wilkins, J.J., Savic, R.M., Karlsson, M.O., Langdon, G., McIlleron, H., Pillai, G., Smith, P.J. & Simonsson, U.S. (2008) Population pharmacokinetics of rifampin in pulmonary tuberculosis patients, including a semimechanistic model to describe variable absorption. Antimicrobial Agents and Chemotherapy, 52, 2138–2148. Zhu, M., Maslow, J.N., Mikota, S.K., Isaza, R., Dunker, F., Riddle, H. & Peloquin, C.A. (2005) Population pharmacokinetics of pyrazinamide in elephants. Journal of Veterinary Pharmacology and Therapeutics, 28, 403–409.
