Population Pharmacokinetics of Diethylcarbamazine in Patients with Lymphatic Filariasis and Healthy Individuals

ABSTRACT

Diethylcarbamazine (DEC) is a drug of choice to treat lymphatic filariasis (LF) either used alone or in combination as mass drug administration (MDA) preventive strategies. The objective of this study was to develop a population pharmacokinetics (PK) model for DEC in subjects infected with lymphatic filariasis (LF) compared to healthy individuals, and to evaluate the effect of covariates on the volume of distribution (V/F) and oral clearance (CL/F) of DEC. This was an open-label cohort study of treatment-naive Wuchereria bancrofti-infected (n = 32) and uninfected (n = 24) adults residing in the Agboville District of Côte d’Ivoire. The population pharmacokinetics model for DEC was built using Phoenix NLME 8.0 software. The covariates included in the model-building process were age, gender, body weight, infection status, creatinine clearance (CL_CR), and aspartate aminotransferase (AST) and alanine aminotransferase (ALT) levels. A total of 56 adults were enrolled in the study, and a total of 728 samples were obtained over 168 h. A one-compartment linear pharmacokinetics model with first-order absorption with an absorption lag time (T_lag) best described the data. After determining the pharmacokinetics (PK) parameters of DEC, body weight and gender were found to be the significant covariates for DEC V/F. The final population pharmacokinetics model adequately described the pharmacokinetics of DEC in the studied population. Model-based simulation indicated that the body weight significantly impacted the exposure in both the male and female populations. This analysis may further support the drug-drug interaction model development of DEC with different coadministered drugs or agents in disease control programs. (This study is registered at clinicaltrials.gov under identifier NCT02845713.)

INTRODUCTION

Lymphatic filariasis (LF), commonly known as elephantiasis, is a neglected tropical disease and caused by infection with nematode parasites of Wuchereria bancrofti, Brugia malayi, or Brugia timori (1). In 2000, 120 million people were estimated to be infected with LF, and now 880 million people in 52 countries remain at risk of acquiring LF (2). LF has therefore been targeted for elimination by the WHO as it is identified as one of the six potentially eradicable infectious diseases identified by the International Task Force for Disease Eradication (3). The WHO recommended mass drug administration (MDA) for LF elimination (4). Diethylcarbamazine citrate (DEC) combined with albendazole (ALB) is the primary drug of choice for MDA to treat LF infection outside sub-Saharan Africa (5). Because DEC plus albendazole fails to sustain microfilarial (Mf) clearance for prolonged periods of time, annual treatment is required for at least 5 years (the approximate life span of adult worms) with sufficient coverage to reduce Mf levels in the community below the threshold to sustain transmission. The addition of ivermectin (IVM) to DEC plus albendazole, or triple-drug therapy, is able to sustain Mf clearance for up to 5 years after a single coadministered regimen (6–8). Consequently, WHO has recommended triple-drug therapy for areas outside sub-Saharan Africa that have failed to eliminate LF with two-drug regimens or in areas that have not started MDA (9). Simulation modeling suggests that only two rounds of triple-drug therapy with sufficient coverage can achieve the LF elimination target in most areas (10).

DEC is thought to be the most potent antifiliarial drug in the triple-drug regimen. It kills both microfilariae and adult worms (11). DEC is cytotoxic to microfilariae within hours after taking the drug; however, significant disappearance of adult worms requires several months to become apparent. DEC is also less efficient at killing adults than microfilariae. A single dose of DEC at 6 mg/kg of body weight inactivates 50 to 80% of adult worms (12–14). Exactly how DEC works to kill microfilariae and adult worms is currently unknown. DEC interferes with parasitic arachidonic acid metabolism via inhibiting prostaglandin H synthase (PGHS) (cyclooxygenase) in the parasites and is thought to make them more susceptible to immune attack (15). Because of the essential role of DEC in MDA for LF outside sub-Saharan Africa and significant pharmacokinetic (PK) variability (16), it is essential to evaluate dosing regimens that produce effective concentrations in diverse populations (i.e., pediatric, obese, etc.).

The study of population pharmacokinetics is an essential aspect of drug development and plays a vital role in finding the optimal and effective treatment dose of a drug of interest (17, 18). Population pharmacokinetics models are efficient in unfolding the pharmacokinetic behavior of investigated drug and to identify the possible source of variation (covariates) that may influence pharmacokinetics (19, 20). The WHO age-based dosing for DEC ranges from 100 to 300 mg across three different age groupings (2 to 5 years, 6 to 15 years, and >15 years) (21). Although age-based dosing is recommended by the WHO for LF MDA programs, the maximum recommended doses using these methods are lower than those indicated by weight-based dosing (6 mg/kg for DEC) and do not consider covariates that could affect drug levels. Using this dosing regimen for DEC may result in significant underdosing of DEC and reduced efficacy in some individuals. There are currently no studies describing the population pharmacokinetics and important covariates of DEC that would help to address this important limitation. The aim of this work is to develop a validated population pharmacokinetics model of DEC and explore the relation between drug levels and patient characteristics (age, sex, and weight) along with infection state and clinical determinates of organ function, including creatinine clearance (CL_CR) and aspartate aminotransferase (AST) and alanine aminotransferase (ALT) levels.

RESULTS

Subject selection, sampling, and noncompartmental pharmacokinetics analysis.

A total of 56 subjects with a Wuchereria bancrofti-infected population (n = 32) and uninfected population (n = 24) were included in the data set. Table 1 summarizes the patient demographics. A total of 728 samples were collected in the study out to 168 h. The plasma DEC concentrations continuously declined after 8 h of administration and were detected only up to 72 h (as DEC was measurable only up to 72 h in all included volunteers) following oral administration of 6 mg/kg (DEC citrate). C_max (1,550 ng/ml), apparent volume of distribution during terminal phase (V_z)/F_obs (114 liters), and CL/F_obs (8.36 liters/h) determined from noncompartmental analysis (NCA) were of further use as initial estimates in development of the population pharmacokinetics model of DEC.

TABLE 1.

Demographics of the study volunteers (n = 56)

Variable	Result for:
	Infected (n = 32)	Uninfected (n = 24)
Age (yr), median (range)	43.50 (22–66)	36.5 (18–54)
Gender, no. (%)
Male	20 (63)	12 (50)
Female	12 (37)	12 (50)
Body wt (kg), median (range)	61.25 (55–135)	63.5 (51–89)
CLCR (ml/min), median (range)	79.80 (41.50–149.0)	79.25 (38.10–137.80)
AST (U/liter), median (range)	29.0 (21–53)	27.0 (15–52)
ALT (U/liter), median (range)	24.50 (14–46)	25.0 (16–67)

Population pharmacokinetics modeling.

On the basis of the goodness-of-fit criteria, DEC concentrations as a function of time were best described by a one-compartment model with an absorption lag time (22). Thus, pharmacokinetic models were parameterized in terms of absorption rate constant (K_a), lag time (T_lag), apparent clearance (CL/F) and apparent volume of distribution (V/F). A one-compartment additive plus proportional error model with T_lag for the given data was developed and compared on the basis of the maximum likelihood criterion (objective function value [OFV] of 2× the negative log likelihood value [−2LL]) and visual inspection of the fittings. The combined additive plus proportional error model with T_lag showed improvement in the OFV (−2LL, 7,635) in comparison to the additive model (−2LL, 8,475) or proportional error model (2LL, 7,765) alone. After adding a T_lag function with combined additive plus proportional error model (base model), OFV values further improved by 304 points (2LL, 7,331). Diagnostic plots for the base model showed a reasonable fit with no apparent trends of residuals over time or model predictions (see Fig. S1 in the supplemental material). The visual covariate screening procedure (scatter plots for continuous and box plots for discrete variables, shown in Fig. S2 and S3 in the supplemental material) indicated the effect of all covariates (age, body weight, gender, infection status, AST, ALT, and CL_CR profile) with PK parameters. To check whether the covariates significantly altered the PK parameters, a model with all the included covariates was run using stepwise forward addition (P ≤ 0.01) followed by a backward elimination (P ≤ 0.001) with the likelihood ratio (−2LL) test. In the forward addition process, the effects of gender, body weight over V/F, and weight over CL/F were included in the model, while in the subtraction process, the effect of weight over CL/F was removed. The covariate analysis clearly indicated the significant effect of weight and gender over apparent volume of distribution (ΔOFV = −40) on DEC pharmacokinetics in this patient population. Eta covariate plots between population parameters and included covariates are shown in Fig. S2 and S3. The final selected population pharmacokinetics model incorporates the effects of body weight and gender covariate on volume (−2LL = 7,291). The final population values for K_a, V/F, CL/F, and T_lag were 1.37 h⁻¹, 103.33 liters, 8.63 liters/h, and 0.74 h, respectively. The percentages of interindividual variability (IIV) calculated from final population model for K_a, V/F, CL/F, and T_lag are 77.45, 7.07, 18.59, and 15.91%, respectively, in comparison to the base model: i.e., 81.51% for K_a, 13.84% for V/F, 19.14% for CL/F, and 15.74% for T_lag. Goodness-of-fit plots for the final DEC population pharmacokinetics model are shown in Fig. 1. The plots of observed plasma concentrations (dependent variable [DV]) of DEC versus individual prediction (IPRED) and population predictions (PRED) were symmetrically distributed, which suggests the good prediction value of the final model. The values of weighted residuals (CWRES) ranged between 2 and −2 and were randomly scattered, indicating the minor differences in the prediction models and the suitability of this error model for this population. The population pharmacokinetics parameters for the final model are shown in Table 2. A summary of the full model development process is shown in Table S1 in the supplemental material in decreasing order of the OFVs depicting model improvement.

FIG 1.

Goodness-of-fit plots of DEC for the final population pharmacokinetics model.

TABLE 2.

Population pharmacokinetics parameter estimates of the final model and bootstrap analysis

Parametera	Final model			Bootstrap
	Estimate	% CV	95% CI	Median	% CV	95% CI
Primary
tvKa (1/h)	1.38	12.96	1.02–1.73	1.37	12.26	1.09–1.77
tvV (liters)	103.33	3.00	97.2–109.4	103.27	2.70	97.69–109.17
tvCL (liters/h)	8.63	2.72	8.17–9.09	8.64	2.57	8.19–9.09
tvTlag (h)	0.74	6.09	0.65–0.83	0.74	6.25	0.62–0.81
tvCmultiSD	0.152	12.61	0.11–0.19	0.151	12.46	0.11–0.19
δVGender1	0.159	23.42	0.08–0.23	0.160	16.53	0.10–0.21
δVwt	0.415	21.99	0.23–0.59	0.405	19.49	0.21–0.53
SD, 0	2.91	26.09	1.42–4.40	3.04	26.70	1.78–5.10
Random effect (% IIV)
ω2Ka	0.47	77.45		0.47	77.45
ω2V	0.005	7.07		0.005	7.07
ω2CL	0.034	18.59		0.033	18.31
ω2Tlag	0.025	15.91		0.03	17.45

^aCV, coefficient of variation; CI, confidence interval; tvK_a, typical value of rate of absorption; tvV, typical value of apparent volume of distribution; tvCL, typical value of apparent clearance; tvT_lag, typical value of absorption lag time; tv_CmultiSD, proportional error on percent scale 0.152 (15.2%); δV_Gender1, exponent for gender as a covariate for volume of distribution; δV_wt, exponent for weight as a covariate for volume of distribution; SD, 0 (square root of CEps variance), additive error on concentration scale (2.91 ng/ml); ω²K_a, variance of interindividual variability for K_a; ω²V, variance of interindividual variability for V; ω²CL, variance of interindividual variability for CL; ω²T_lag, variance of interindividual variability for T_lag.

Model evaluation.

The stability, precision, and robustness of the final model were assessed using nonparametric bootstrap analysis. Table 2 provides comparison of the final model parameter estimates and bootstrap replicates (n = 1,000) for DEC. The parameter estimates of the final population pharmacokinetics model were close to the median values obtained from bootstrap replications, showing the robustness, accuracy, and stability of the final population PK model. The low percentage of coefficient of variation (CV) for the bootstrap results suggests parameter uncertainty was small. The bootstrap calculated percentage of IIV was similar to the final model estimated for the patient‐to‐patient variability. Visual predictive check (VPC) plots were constructed to evaluate the model predictability. Concentration-time data were graphically superimposed on the median values and the 5th and 95th percentiles of the simulated concentration-time profiles. In addition, The VPC (n = 1,000) results for the final model presented in Fig. 2 showed that the most of observed DEC data fell within the boundaries of 5th and 95th percentiles, suggesting the adequacy of the final model to explain the observed data. Normalized prediction distribution error (NPDE) was also evaluated. The VPC and bootstrap parameter values both showed that the model was a good description of the data (Table 2 and Fig. 2).

FIG 2.

The visual predictive check of the final population pharmacokinetics model over the time (independent variable [IVAR]) from 0 and 72 h following DEC administration (n = 1,000) is shown. The inset shows the first 20 h after dosing.

Using the final model, simulations of plasma-concentration profiles were performed for DEC administered at a dose of 6 mg/kg. Descriptive statistics of AUC_last (area under the plasma-concentration curve from dosing to 72 h postdose) and C_max are provided in Table 3.

TABLE 3.

Simulated AUC_last (0 to 72 h) and C_max values after 6 mg/kg of DEC

Parameter	Median	Mean	SD	95% CI
AUClast (0–72 h) (ng · h/ml)	22,755	23,109	5,012.2	13064.7–33154.3
Cmax (ng/ml)	1,570.2	1,597.4	241.6	1113.1–2081.7

C_max and AUC_last for the 6-mg/kg dose were simulated for a male subject and female subject (age of 40 years) with a weight of 51 kg (minimum-weight person as per the given population), an average-weight person (65 kg), and an overweight person (135 kg) to assess the impact of gender along with weight on drug exposure.

A lower-weight female (51 kg) had 26% lower C_max and AUC_last values than the average-weight female (65 kg [average weight as per the studied population]), while an overweight female (135 kg) had 10% lower C_max and 50% higher AUC_last values than the average-weight female. On the other hand, A lower-weight male (51 kg) had 7% lower C_max and 8% lower AUC_last values than the average-weight male (65 kg), while an overweight male (135 kg) had 10% lower C_max and 37% higher AUC_last values than an average-weight male.

In comparison of the female to male population, a lower-weight female had 5% higher C_max and 27% lower AUC_last values, an average-weight female had 33% higher C_max and 3% lower AUC_last values, and an overweight female had 8% higher C_max and 6% lower AUC_last values.

DISCUSSION

The objective of this study was to describe the population PK parameters of DEC administered orally at 6 mg/kg in clinical patients infected with LF caused by Wuchereria bancrofti and uninfected healthy adults. Blood samples for drug levels were obtained at 1, 2, 3, 4, 6, 8, 12, 24, 36, 48, 72, and 168 h after treatment. This is the first population-based assessment of the pharmacokinetics of DEC in patients and healthy volunteers. Population modeling offers an opportunity to assess drug disposition properties taking into account individual’s pharmacokinetic variability. The pharmacokinetics data of DEC are well described by a one-compartment linear model with first-order absorption. We have considered age, weight, infection status, gender, AST, ALT, and CL_CR as covariates to be included in the model development process. We have not included effect of coadministration of IVM and ALB on DEC pharmacokinetics as the findings of the prior study (7, 23) show no interaction between these drugs. We have previously shown that the triple-drug combination is also safe in uninfected individuals and can therefore be used in MDA programs (16). The combined additive-proportional residual error model with T_lag function was found to best describe the data. In this study, the inclusion of age and health condition/infected state in the base model did not appear to improve the data fitting. Among the considered covariates, a correlation between body weight and gender over volume of distribution was observed (OFV = −40). A previous finding described that intersubject variations in the pharmacokinetics parameters of DEC could be due to nonadjustment of dose to body weight (24). However, our DEC dosing was dependent on body weight, but to a certain degree. The unavailability of ideal body weight (due to lack of data) serves as a limitation of the generated model to clearly defining the gender and body weight correlation. In addition, this study only included adult subjects (18 to 66 years), which makes extrapolation of dosing into pediatric and elderly patients difficult. Goss and coworkers developed dosing pole recommendations for DEC to compare the current WHO’s height- or age-based dosing recommendation as alternatives to weight-based dosing for mass drug administration lymphatic filariasis (LF) elimination programs. Their results supported weight-based DEC dosing along with other factors affecting height-weight relationships (e.g., age, sex, and geographic location) (25).

The CWRES distribution of the final model revealed uniform distribution, and the value ranged between 2 and −2, indicating the slight differences in prediction models. As part of a validation process, the bootstrap has been run for a 1,000-data set random sampling of the subjects from the original data set, and the results were in line with our final population PK model. VPC (n = 1,000) also explains the adequacy of our final model. Finally, our study confirms the significant relationship between V/F and body weight and gender for DEC. Further simulation suggests that the average-weight female population had higher C_max values than average-weight male subjects, but as the gender covariate impacted volume of distribution, the AUC of exposure was not changed significantly. Body weight significantly impacted the exposure (increased body weight increased exposure) in both the male and female populations. This effect was more pronounced in the male population. However, this population PK study was performed in a specific population in West Africa, and there may be limitations applying these findings globally.

Conclusion.

We developed a population pharmacokinetic model of DEC in Wuchereria bancrofti-infected (n = 32) and uninfected (n = 24) populations. In this study, nonlinear mixed-effect modeling (NLME) was performed. The final model showed a significant influence of body weight and gender on volume of distribution of DEC. The stability, robustness, and the predictive performance of the final population PK model were established by using goodness-of-fit plots, bootstrapping, and VPC. The developed model along with described simulations may guide future study designs and be useful for predicting pharmacokinetic/pharmacodynamic analyses.

MATERIALS AND METHODS

Subjects and study design.

This was an open-label cohort study of treatment-naive Wuchereria bancrofti-infected (n = 32) and uninfected (n = 24) adults residing in Agboville District of Côte d’Ivoire. LF is endemic in the Agboville District, and onchocerciasis is nonendemic. Informed consent was obtained from all participants. Individuals (18 to 70 years) included in this study were prescreened for W. bancrofti infection with fingerstick blood using a filariasis test strip (FTS; Alere, Inc., Waltham, MA, USA), to detect circulating filarial antigenemia. Exclusion criteria for the study were a positive pregnancy test, chronic kidney or liver disease, or a serum alanine transaminase, aspartate transaminase, or creatinine level >1.5 times the normal limits or severe anemia (blood hemoglobin, <7 g/dl), taking medication that could interfere with test drug metabolism within 1 week of study onset, evidence of urinary tract infection (as indicated by an active urinary sediment of >68 pus/neutrophil cells per field or 3+ nitrate on dipstick), or lactose and/or gluten intolerance. All individuals were tested for onchocerciasis by microscopic examination of two skin snips taken from the iliac crests and antibodies to a recombinant Onchocerca volvulus antigen (Ov16 rapid test; Standard Diagnostics, Inc., Youngin, South Korea) since DEC can cause serious adverse events in persons with onchocerciasis (26). Participants were given a single coadministered dose of IVM (200 μg/kg), DEC citrate (6 mg/kg), and ALB (400 mg) as directly observed therapy. A peripheral intravenous catheter was placed for the first 12 h due to frequent blood draws. Blood samples for drug concentrations were obtained prior to and at 1, 2, 3, 4, 6, 8, 12, 24, 36, 48, 72, and 168 h after treatment, and plasma was separated by centrifugation and stored at –20°C. Institutional review boards in Cleveland, OH (University Hospitals Cleveland Medical Center IRB no. 03-16-09), and in Côte d’Ivoire approved the study protocol (Comité National d’Ethique et de la Recherche, CNER, N/Ref: 022/MSLS/CNER-kp). This trial is registered at clinicaltrials.gov (identifier NCT02845713).

Bioanalytical assays.

Plasma concentrations of DEC were determined using a validated liquid chromatography-tandem mass spectrometry (LC-MS/MS) method as previously described (22). The assay was linear from 1 to 2,000 ng/ml. Control concentrations of 5, 500, and 1,500 ng/ml were within 15% of nominal values for all data.

Population pharmacokinetics modeling.

The concentration-time data of DEC was used to develop the population pharmacokinetics model by using a nonlinear mixed‐effects approach with Phoenix NLME software (version 8.0; Certara L.P., St. Louis, MO, USA). The first-order conditional estimation (FOCE) method was used throughout the model-building process. One- and two-compartment models with and without lag time were evaluated with different residual errors to identify the model that best described the set of data. Initially, the base structural model without covariates was generated by using the initial estimates (observed C_max, V_z/F_obs, and CL/F_obs of noncompartmental analysis [NCA]). The best structural model was chosen on the basis of examination of objective function value (OFV; equal to the twice the negative log likelihood [−2LL] value), the Akaike information criterion (AIC), and the visual inspection of diagnostic standard goodness-of-fit plots (16, 17). The structural pharmacokinetics base model comprised of three fundamental fixed‐effect parameters: apparent clearance (CL/F), apparent volume of distribution (V/F), and the absorption rate constant (K_a).

The interindividual variability of the PK parameters was modeled by the log normal distribution of the exponential model as shown in equation 1:

$$V=\text{tv}\times \exp (\text{η})$$ (1)

Here, V, CL, and K_a are the fixed effects for the pharmacokinetic parameter, tv represents the typical value of the population mean (tvV, tvCL, and tvK_a), and η (ηV, ηCL, and ηK_a) is a random variable following a Gaussian distribution with a mean of 0 and a variance of ω². The interindividual variation, η, for between-subject variability and the interindividual variation for random variables with high shrinkage value (>0.5) were not included in the model (27). The additive, proportional, and combined additive plus proportional error models were tested, and the residual error model was best described using a combined additive-proportional error model with T_lag function as shown in equation 2:

$$C_{\text{obs}}=C+C_{\text{Eps}}\times \text{sqrt}\left(1+\left(C^2\times {\left(\frac{C_{\text{multiSD}}}{\text{sigma}()}\right)}^2\right)\right)$$ (2)

C_obs and C are the observed and predicted concentrations of the individual, respectively, C_Eps is a normally distributed residual error with a mean of 0 and ω² (additive part of variance), sqrt, is the square root, C_multiSD is C with multiple standard deviations, and C² × (C_multiSD/sigma())²) is proportional part variance.

Population covariate analysis.

In the second step of model building, demographic variables were tested as potential covariates for the PK parameters. This study included sex and infection state as categorical covariates and age, weight, CL_CR, and AST and ALT levels as continuous covariates. CL_CR was calculated from serum creatinine data using the following equation:

$${\text{CL}}_{\text{CR}}=(140-\text{age\hspace{0.17em}[yr]\hspace{0.17em}}\times \text{\hspace{0.17em}wt\hspace{0.17em}[kg}])/(\text{72\hspace{0.17em}}\times \text{\hspace{0.17em}serum\hspace{0.17em}CR\hspace{0.17em}[mg/dl])}$$ (3)

(Multiply the result by 0.85 for the female.)

A visual covariate screening procedure was performed before modeling, which includes scatter plots for continuous covariates and box plots for discrete variables. After visual inspection, all the covariates were added, and the model with covariates was run using the stepwise forward addition procedure (P ≤ 0.01) followed by a backward elimination procedure (P ≤ 0.001) with the likelihood ratio test. All of the selected covariates were evaluated according to the difference in the OFV (−2LL), AIC, and further change in interindividual variability (percentage of IIV) among individual PK parameters and covariate plots. A covariate was considered significant when the addition of the covariate resulted in a decrease in the OFV greater than 6.63 (P < 0.01), and the elimination of the covariate resulted in an increase in the OFV greater than 10.82 (P < 0.001) (28). The percentage of IIV was calculated by using the following equation:

$$\text{\%\hspace{0.17em}IIV}=\sqrt{e({\text{ω}}^2)-1}\times 100$$ (4)

After a full covariate regressive model was established, the analysis procedure was continued until only significant covariates remained in the model. At the end of the model‐building process, all significant covariates were then incorporated into the base model, which was designated the “final” population model.

Model validation and model-based simulations.

The suitability of the final full population model was evaluated using the nonparametric bootstrap analysis and visual predictive check (VPC) to evaluate the precision and robustness of the final model (29). A bootstrap (n = 1,000) was performed by random sampling of the subjects from the original data set. The final parameter estimates obtained from the original data set were compared with the median values and 95% confidence intervals (CIs) of the bootstrap estimates. The final population model of DEC was evaluated on the basis of a visual predictive check using a 1,000-data set simulation to determine whether the fitted model provides an adequate description of the data. The lower (5%), median (50%), and upper (95%) percentiles of the simulated concentrations were calculated and then visually checked to evaluate how the simulated results are related to the observed DEC concentrations.

Using the parameter estimates from the final model, plasma drug concentration profiles were simulated for the given doses 6 mg/kg, using the same population included in the population pharmacokinetics modeling regarding covariate distributions. One thousand plasma drug concentration profiles per dose were simulated. From the plasma drugconcentration profiles and estimated CL/F values, the pharmacokinetics parameters AUC_last (area under the plasma-concentration curve from dosing to 72 h postdose) was estimated as dose/(CL/F) and C_max (maximal plasma drug concentration) was read directly from the plasma drug concentration profiles. The arithmetic mean, median, standard deviation (SD), and the 2.5% and 97.5% percentiles for AUC_last and C_max were estimated.