Abstract
Theophylline is commonly used for the treatment of asthma and has a low hepatic clearance. The changes in plasma albumin concentration occurring in asthma may affect the exposure of theophylline. The aim of the presented work was to predict theophylline pharmacokinetics (PK) after incorporating the changes in plasma albumin concentration occurring in patients with asthma into a physiologically based pharmacokinetic (PBPK) model to see whether these changes can affect the systemic theophylline concentrations in asthma. The PBPK model was developed following a systematic model building approach using Simcyp. The predictions were performed initially in healthy adults after intravenous and oral drug administration. Only when the developed adult PBPK model had adequately predicted theophylline PK in healthy adults, the changes in plasma albumin concentrations were incorporated into the model for predicting drug exposure in patients with asthma. After evaluation of the developed model in the adult population, it was scaled to children on physiologic basis. The model evaluation was performed by using visual predictive checks and comparison of ratio of observed and predicted (Robs/Pre) PK parameters along with their 2-fold error range. The developed PBPK model has effectively described theophylline PK in both healthy and diseased populations, as Robs/Pre for all the PK parameters were within the 2-fold error limit. The predictions in patients with asthma showed that there were no significant changes in PK parameters after incorporating the changes in serum albumin concentration. The mechanistic nature of the developed asthma-PBPK model can facilitate its extension to other drugs.
SIGNIFICANCE STATEMENT Exposure of a low hepatic clearance drug like theophylline may be susceptible to plasma albumin concentration changes that occur in asthma. These changes in systemic albumin concentrations can be incorporated into a physiologically based pharmacokinetic model to predict theophylline pharmacokinetics in adult and pediatric asthma populations. The presented work is focused on predicting theophylline absorption, distribution, metabolism, and elimination in adult and pediatric asthma populations after incorporating reported changes in serum albumin concentrations to see their impact on the systemic theophylline concentrations.
Introduction
Asthma is a chronic disorder that affects people from all age groups in the whole world (Currie et al., 2005; Global Institute for Asthma, 2018). Asthma is characterized by variable symptoms that can be managed by initiating appropriate drug therapy (Ferraro et al., 2018). Since asthma is a disease that requires long-term management, the selection of drug and its dose has to made very carefully after taking into account all the possible drug-drug and drug-disease interactions (Taburet and Schmit, 1994). Asthma is reported to be associated with changes in serum albumin concentrations; these pathophysiological changes in serum albumin levels can potentially affect the absorption, distribution, metabolism, and elimination (ADME) of administered drugs (Mitenko and Ogilvie, 1973). The changes in serum albumin concentration occurring in asthma can potentially affect the unbound drug concentrations, particularly for the drugs with low hepatic clearance. Therefore, the changes in serum albumin levels should be considered while describing the pharmacokinetics (PK) of drugs being administered in asthma (Blanchard et al., 1992; Shima and Adachi, 1996; Picado et al., 1999; Vural and Uzun, 2000; Misso et al., 2005; Ejaz et al., 2017).
The physiologically based pharmacokinetic (PBPK) approach provides novel opportunities to incorporate the relevant pathophysiological changes occurring in various diseases for the construction of disease models (Li et al., 2012; Park et al., 2017). There are some published examples of drug-disease PBPK models that incorporate pathophysiological modifications occurring in different chronic conditions (Edginton and Willmann, 2008; Johnson et al., 2010; Rowland Yeo et al., 2011; Li et al., 2012, 2015; Schaller et al., 2013; Sayama et al., 2014; Vogt, 2014; Chadha and Morris, 2015; Rasool et al., 2016; Shah et al., 2019). Although there are a few published reports of PBPK models for theophylline in adults and children (Ginsberg et al., 2004; Björkman, 2005), until now there has been no published report of a theophylline-asthma PBPK model that has been used to predict drug exposure in adult and pediatric patients with asthma after incorporation of changes in human serum albumin concentrations. Therefore, if such a PBPK drug-disease model is developed, it may have many clinical implications.
Theophylline is a low hepatic extraction drug that is commonly used for the treatment of asthma (Mazza, 1982). It has a narrow therapeutic index and is usually reserved for those patients with asthma who cannot be treated with conventional combination drug therapies (Kim and Mazza, 2011). Theophylline belongs to Biopharmaceutical Classification System class I, having high solubility and permeability; this is why it is rapidly absorbed after oral administration and has a bioavailability range of 80%–100% (Griffin et al., 2013; Taburet and Schmit, 1994). It undergoes hepatic metabolism through different cytochrome P450 (P450) enzymes (CYP1A2, CYP2E1, and CYP3A4) (Kim and Mazza, 2011). The availability of clinical PK data for theophylline in the published literature makes it an ideal candidate for development and evaluation of its PBPK model in adult and pediatric patients with asthma (Mitenko and Ogilvie, 1973; Ellis et al., 1976; Richer et al., 1982; Hendeles and Weinberger, 1985; Björkman, 2005). It is already known that plasma protein binding may have a significant impact on the exposure of low hepatic extraction drugs like theophylline (Colli et al., 1988; DiPiro, 2010). Therefore, if a theophylline-asthma PBPK model is developed that incorporates the plasma albumin changes, it can be used to predict the systemic theophylline concentrations in patients with asthma. Furthermore, after evaluation of the developed theophylline-asthma PBPK model in adults, it can be scaled to pediatric patients with asthma on a physiologic basis by using a PBPK simulator.
The aim of the presented work was to predict theophylline PK in adults and children with asthma after incorporating the changes in plasma albumin concentration into a PBPK model to see whether these changes can affect the systemic theophylline concentrations in asthma.
Materials and Methods
Modeling Software
The PBPK model was developed using Simcyp population-based simulator, version 16, release 1 (Certara UK Limited, Simcyp Division, Sheffield, UK).
Strategy for Model Building
A total of 24 PK profiles from 16 clinical studies were selected for model development and evaluation, which includes 17 in healthy adults (intravenous, 6; oral, 11), four in adult patients with asthma (intravenous, one; oral, three), and three (intravenous, one; oral, two) in pediatric patients with asthma. The details of the population data used for model development and evaluation can be seen under the heading Clinical/Pharmacokinetic Data. The development and verification of the PBPK model was based on previously reported systematic model building and verification approaches (Khalil and Läer, 2014; Rasool et al., 2015; Sager et al., 2015). The systematic model building approach is focused on the selection and optimization of drug-specific model input parameters that are responsible for predicting drug disposition in healthy adults after intravenous administration. After evaluating the developed model with the intravenous clinical PK data, the parameters that govern the oral drug absorption process are selected and optimized for predicting the drug PK after oral application. This was done by using 33% of clinical PK data sets (n = 6, three intravenous and three oral) in adults for model parameterization and the remaining 66% (n = 11) for subsequent model verification. All these clinical PK data sets were included in final model evaluation. After model evaluation in healthy adults, the disease-specific parameters were incorporated into the developed model to constitute the disease model. Thereafter, the developed PBPK disease model was used to predict ADME of administered drug in the diseased population. After the evaluation of the developed PBPK models in adult healthy and diseased populations, it was scaled to children on physiologic basis by using the Simcyp pediatric module. The implemented model building strategy can be seen in Supplemental Fig. 1.
Model Structure and Parameterization
To identify and select various model input parameters, a detailed literature review was conducted. The input parameters were selected from the Simcyp theophylline compound library and from the published literature. The final model input parameters can be seen in Supplemental Table 1. A detailed explanation of model parameterization is given below.
Absorption
The advance dissolution, absorption, and metabolism (ADAM) model was used for prediction of oral drug absorption process. This model incorporates information on various physiologic factors that can influence drug absorption process, such as gastric emptying time, small intestine transit time, gastrointestinal pH, fluid dynamics, abundance of gut wall enzymes, transporters, and segregated segmental blood flows (Jamei et al., 2009). The reported theophylline human jejunum permeability (Peff,man) of 4.2 × 10−4 cm/s was used for prediction of oral drug absorption process (Simcyp compound library, version 16, release 1). The predicted value of absorbed drug fraction (fa) was 0.93; that is in line with the reported literature, as theophylline belongs to Biopharmaceutical Classification System class I and has a high permeability and a high solubility (Chavda et al., 2010). It is known that the gastric emptying time varies with the type of administered drug formulation. The oral drug formulations like pellets are reported to be absorbed in a different manner when compared with the conventional oral dosage formulations (Davis et al., 1986). To predict the absorption of oral theophylline pellets, their dissolution profile was incorporated into the ADAM model (Gonzalez and Golub, 1983).
Distribution
Minimal PBPK model was used for the prediction of drug distribution. The steady state volume of distribution (Vss) and tissue-plasma partition coefficient (Kp) were predicted by using the Poulin and Theil method (Poulin and Theil, 2002). The predicted and reported values for Vss were 0.45 and 0.51 l/kg (Obach et al., 2008).
Elimination
Theophylline is metabolized in liver via CYP1A2, CYP2D6, CYP2E1, and CYP3A4 enzymes (Ha et al., 1995). The individual values of maximum rate of reaction (Vmax) and Michaelis-Menten constant (Km) for each P450 enzyme were used for predicting theophylline metabolism (Simcyp compound library, version 16, release 1). A renal clearance value of 0.31 l/kg was also used for predicting theophylline elimination (Simcyp compound library, version 16, release 1).
Disease Model
It is known that the systemic concentrations of drugs with low hepatic clearance may be susceptible to changes in their serum albumin concentration. Therefore, any pathophysiological condition that can potentially affect the serum albumin concentration of low clearance drugs will affect their systemic concentrations (both unbound and bound) and pharmacodynamic effect. Moreover, these serum albumin concentration changes may cause serious therapeutic problems (adverse drug reactions and toxicity) if a low clearance drug with a narrow therapeutic index is administered (Funk et al., 2012).
It has been reported that human serum albumin (HSA) concentrations are altered in patients with asthma (Blanchard et al., 1992; Shima and Adachi, 1996; Picado et al., 1999; Vural and Uzun, 2000; Misso et al., 2005; Ejaz et al., 2017). The serum albumin concentration is reduced to 41 g/l in patients with asthma (Blanchard et al., 1992). To predict systemic drug concentrations in patients with asthma, the model input value of serum albumin concentration was reduced from 50.34 g/l in male patients and 49.38 g/l in female patients to 41 g/l (similar in both male and female patients) within the Simcyp population library. These reductions in serum albumin concentrations were incorporated into the developed PBPK model to predict theophylline exposure in adult asthma population.
Model Scaling to Children
When the adult model had adequately predicted theophylline ADME in adult healthy and diseased populations, it was scaled to children by using Simcyp pediatric module. The Simcyp pediatric population incorporates the age specific anatomic and physiologic changes that form the basis for PK differences between adult and pediatric populations (Johnson and Rostami-Hodjegan, 2011; Johnson et al., 2018). All the pediatric patients included in model evaluation were having asthma. The main enzymes involved in the metabolism of theophylline are CYP3A4, CYP2E1, CYP2D6, and CYP1A2. The differences in ontogeny profiles of these P450 enzymes may lead to age-related changes in theophylline clearance (Salem et al., 2013). Due to the absence of information on serum albumin changes in pediatric patients with asthma, the incorporated changes in serum albumin concentrations in the developed pediatric model were based on adult values.
Clinical/Pharmacokinetic Data
In Adults.
To search, screen, and extract the clinical PK data on theophylline in healthy and diseased populations, a comprehensive literature review was performed using the online search engines Google Scholar and PubMed. Initial screening of PK studies was based on the presence of systemic concentration versus time profiles of theophylline in the published literature. The final selection of clinical PK data was based on the presence of clear information on age, weight, administered dose (oral and intravenous), disease condition, the proportion of female patients, and the fasting/fed state. Finally, nine pharmacokinetic studies were selected for model development and evaluation in healthy adults. These studies include 17 systemic concentration versus time profiles (intravenous, 6; oral, 11) in 114 healthy individuals (Chrzanowski et al., 1977; Ishizaki et al., 1979; Antal et al., 1981; Rovei et al., 1982; Gonzalez and Golub, 1983; Gundert-Remy et al., 1983; Lagas and Jonkman, 1983; St-Pierre et al., 1985; Lelo et al., 1986). Characteristics of healthy population data used for theophylline model development are given in Supplemental Table 2. The clinical PK data used for model evaluation were based on reported mean systemic concentration versus time profiles (Chrzanowski et al., 1977; Ishizaki et al., 1979; Antal et al., 1981; Rovei et al., 1982; Gonzalez and Golub, 1983; Gundert-Remy et al., 1983; St-Pierre et al., 1985; Lelo et al., 1986) except in one study where individual systemic concentration versus time profile was used for comparison (Lagas and Jonkman, 1983). Four studies were selected in patients with asthma with four mean PK profiles (intravenous, two; oral, two) in 134 adults with asthma (Mitenko and Ogilvie, 1973; Richer et al., 1982; Steinijans et al., 1982; Weinberger and Hendeles, 1986). Characteristics of data used for theophylline model development in adult asthma population are given in Supplemental Table 3. The observed data were extracted by scanning the drug concentration versus time graphs from literature using Get Data Graph Digitizer software (version 2.26).
In Children.
Three clinical PK studies of theophylline (intravenous, two; oral, one) with 62 children having asthma were included for model evaluation in the pediatric population (Ellis et al., 1976; Loughnan et al., 1976; Hendeles and Weinberger, 1985). Characteristics of clinical PK data used for theophylline model development in asthma pediatric population are given in Supplemental Table 3.
Model Evaluation
Simulations were performed by creating a virtual population of 100 individuals (10 trials of 10 individuals) with similar demographic characteristics (age range, proportion of female patients, fasting/fed state) as in reported clinical studies. The visual predictive checks (VPCs) were used for initial model evaluation. In VPCs, the mean observed and predicted systemic drug concentration versus time plots were overlaid for direct visual comparison. Additionally, the 5th–95th percentiles and maximum and minimum predictions were also used for model evaluation.
A noncompartmental analysis was performed for the comparison of observed and predicted PK parameters, such as the area under the systemic drug concentration-time curve from time zero to last measured systemic drug concentration (AUC0-last), the maximum concentration of drug (Cmax), clearance of drug (CL; CL/F after oral application) by using the Excel add-in program pK Solver (Zhang et al., 2010). Furthermore, the ratio of observed and predicted (Robs/Pre) PK parameters like Cmax, AUC0–last, and CL was also calculated along with 95% confidence intervals (CIs) for model evaluation (eq. 1). A 2-fold error range was used as a reference for all the evaluations (De Buck et al., 2007; Li et al., 2012; Khalil and Läer, 2014). Additionally, fold error, average fold error (AFE), and root mean square error (RMSE) were also used for assessing model accuracy and precision (eqs. 2–4):
Ratio (Robs/pre)(1)(2)(3)(4)
Results
Healthy Adult Population
Intravenous Doses.
The observed and predicted systemic theophylline concentration versus time profiles after administering different intravenous doses (4–6 mg/kg and 193.2–386.4 mg) in the healthy population are shown in Fig. 1, A–F (Chrzanowski et al., 1977; Ishizaki et al., 1979; Gundert-Remy et al., 1983; St-Pierre et al., 1985). It is apparent from the VPCs that the model has successfully apprehended the observed PK data after intravenous application. The mean RObs/Pre for AUC0–last was 0.89 (95% CI 0.82–0.97), and the mean RObs/Pre values for Cmax and CL were also within the acceptable 2-fold error range (Fig. 3; Table 1). Additionally, residual plots demonstrated that there was no systematic error in model prediction (Supplemental Fig. 2, A–C). Furthermore, the AFE and RMSE values showed that the developed model has adequately described theophylline PK after intravenous application (Supplemental Table 4).
Oral Doses.
The observed and predicted systemic theophylline concentration versus time profiles after application of different oral doses of theophylline, i.e., 125–600 mg in the healthy population, are shown in Fig. 1, G–P (Antal et al., 1981; Rovei et al., 1982; Gonzalez and Golub, 1983; Lagas and Jonkman, 1983; Lelo et al., 1986). It can be seen from the comparison of observed and predicted systemic concentrations through VPCs that the model has successfully predicted theophylline PK after oral application. The mean AUC0-last RObs/Pre value after oral application was 1.05 (95% CI 0.92–1.18), and the mean RObs/Pre values for Cmax and CL were also within the acceptable 2-fold error range (Fig. 3; Table 1). Moreover, the residual plots showed that all the predicted concentrations were in harmony with the observed data and there was no systematic error in model predictions (Supplemental Fig. 2, D–F). Additionally, the AFE and RMSE values for all the PK parameters showed that the model had successfully predicted these parameters (Supplemental Table 4).
Adult Patients with Asthma
The developed asthma model has efficiently predicted systemic theophylline concentrations after administration of both intravenous (5.6 mg/kg and 351 mg) and oral (7.5 mg/kg and 600 mg) doses in patients with asthma (Fig. 2, A–D) (Mitenko and Ogilvie, 1973; Richer et al., 1982; Steinijans et al., 1982; Weinberger and Hendeles, 1986). The VPCs (Fig. 2, A–D) and residual plots (Supplemental Fig. 2, G–L) showed that the model predictions were in agreement with the observed clinical PK data. However, after analyzing the predictions closely, it was seen that the incorporation of the reported decrease in HSA resulted in no significant changes, as the mean CL RObs/Pre after intravenous and oral application with HSA changes were 0.93 and 1.04 compared with the mean CL RObs/Pre of 1.02 and 1.08 without incorporation of changes in HSA. The mean RObs/Pre values for AUC0-last and Cmax were within the 2-fold error range (Fig. 3, A, C, and E; Table 2). Furthermore, the AFE and RMSE values showed that the model had adequately predicted theophylline PK in patients with asthma (Supplemental Table 4).
Pediatric Patients with Asthma
The predicted systemic theophylline concentrations after intravenous (3.2–4 mg/kg) and oral (8.2 mg/kg) administration in children with asthma were in agreement with the observed data (Fig. 2, E–G) (Ellis et al., 1976; Loughnan et al., 1976; Hendeles and Weinberger, 1985); this was further confirmed by looking into residual plots (Supplemental Fig. 3, A–F). Moreover, the mean CL RObs/Pre with and without considering changes in HSA after intravenous and oral theophylline administration were 1.5 and 1.0, respectively. The mean RObs/Pre for AUC0-last and Cmax was within the 2-fold error range (Fig. 3, B, D, and F; Table 2). Additionally, the AFE and RMSE values in children with asthma showed that the model had predicted theophylline PK adequately (Supplemental Table 4).
Discussion
In this study, the PBPK approach was used for the prediction of theophylline PK in healthy and diseased populations (adult and pediatric). The PBPK model development process was initiated by selecting drug specific parameters that govern theophylline disposition in healthy adults after intravenous administration. When the developed PBPK model was successfully evaluated with the reported clinical PK data, the parameters that control the oral drug absorption process were incorporated into the model for predicting ADME of theophylline in healthy individuals after oral application. Once the developed PBPK model had successfully predicted theophylline PK in healthy adults, the reported pathophysiological changes in HSA that occur in patients with asthma were incorporated into the model for predicting theophylline ADME. Only when the developed adult PBPK model had adequately predicted theophylline ADME in healthy and diseased populations was the model scaled to children by using the Simcyp pediatric module. The model predictions in patients with asthma showed that there were no significant changes in PK parameters after incorporating the reported pathophysiological changes in HSA. The developed PBPK model has effectively described theophylline PK in both healthy and diseased populations.
The developed model has successfully predicted systemic theophylline concentrations in healthy population after intravenous drug administration, which is evident from the agreement between the mean observed and predicted CL values of 0.05 l/h per kilogram (95% CI 0.03–0.06) and 0.04 l/h per kilogram (Chrzanowski et al., 1977; Ishizaki et al., 1979; Gundert-Remy et al., 1983; St-Pierre et al., 1985). Similarly, the mean observed and predicted CL/F values after oral administration of theophylline were 2.73 l/h (95% CI 2.10–3.35) and 2.72 l/h (95% CI 2.38–3.06) (Antal et al., 1981; Rovei et al., 1982; Gonzalez and Golub, 1983; Lagas and Jonkman, 1983; Lelo et al., 1986). The predicted oral bioavailability of 93% (62%–99%) was within the reported range of 80%–100% (Griffin et al., 2013; Taburet and Schmit, 1994). Moreover, the mean observed and predicted AUC0-last and Cmax after intravenous and oral application were also within the 2-fold error range. The values of AFE and RMSE after intravenous and oral application suggested that the developed model has effectively described the PK of theophylline.
Theophylline is a drug with low hepatic clearance and a narrow therapeutic index. It is already known that exposure of drugs with low hepatic clearance is susceptible to changes in plasma protein binding (Zhivkova, 2017). Since theophylline is a low hepatic clearance drug that is bound to albumin, any change in serum albumin concentration can potentially affect its ADME and systemic concentrations (Fleetham et al., 1981; Blanchard et al., 1992). Moreover, it is reported that asthma is associated with pathophysiological reduction in serum albumin concentration and that these reductions in HSA levels may affect serum concentration of theophylline in patients with asthma (Blanchard et al., 1992). Keeping in view the reported pathophysiological reductions in HSA, its input value was reduced to 41 g/l in the developed model for predicting theophylline ADME in patients with asthma (Blanchard et al., 1992). The developed model has successfully predicted theophylline CL in adult patients with asthma after administering intravenous doses of 5.6 mg/kg and 351 mg, as the observed theophylline CL was 0.04 l/h per kilogram and 2.41 l/h, whereas the predicted CL was 0.05 l/h per kilogram and 3.21 l/h, respectively (Mitenko and Ogilvie, 1973; Steinijans et al., 1982). Similarly, the predicted theophylline CL/F after administering oral doses of 600 mg and 7.5 mg/kg was 2.70 l/h and 0.04 l/h per kilogram, which was comparable to the observed values of 2.93 l/h and 0.04 l/h per kilogram (Richer et al., 1982; Weinberger and Hendeles, 1986).
To see the impact of plasma protein binding changes on ADME of theophylline in patients with asthma, the simulations were performed with and without incorporation of changes in HSA. There were no significant differences seen in predicted PK parameters after comparing the simulations with and without the incorporation of changes in HSA. Moreover, the unbound theophylline concentration was also predicted to see whether the changes in HSA had any significant effect on the unbound AUC0-last in adults with asthma. There were minor differences seen in the predicted unbound AUC0-last after intravenous and oral administration in adult patients with asthma with and without incorporation of changes in HSA. The unbound AUC0-last in adults with asthma after administering 5.6 mg/kg intravenous theophylline with and without changing HSA was 46.1 and 44.5 mg/l.h. Similarly, after administering 600 mg oral theophylline the unbound AUC0-last with and without changing HSA was 102.8 and 100 mg/l.h. The probable reason behind not seeing any significant change in predictions after incorporating changes in HSA in the patients with asthma may be supported by the fact that changes in plasma protein binding have minimal effect on the exposure of low clearance drugs (Rowland, 1984; Benet and Hoener, 2002; Heuberger et al., 2013).
The developed adult PBPK model after the evaluation was scaled to children on physiologic basis by using the pediatric module of Simcyp. Since there was no information available in the published literature regarding the changes in HSA in pediatric patients with asthma, the pathophysiological reductions in HSA that were used in the adult disease model were adopted in the pediatric model. The developed pediatric PBPK model successfully predicted theophylline CL in pediatric patients with asthma after administering intravenous doses of 3.2 and 4 mg/kg, as the observed theophylline CL was 0.11 l/h per kilogram, whereas the predicted CL was 0.07 l/h per kilogram (Ellis et al., 1976; Loughnan et al., 1976). Similarly, the observed and predicted theophylline CL/F after administering oral doses of 8.2 mg/kg in children with asthma was 0.07 l/h per kilogram (Hendeles and Weinberger, 1985). In comparison with adults, the observed theophylline CL is higher in children with asthma (5.6 mg/kg intravenous: adults 0.04 l/h per kilogram vs. 4 mg/kg intravenous children: 0.11 l/h per kilogram; after oral: 7.5 mg/kg adults 0.04 l/h per kilogram vs. 8.2 mg/kg children 0.07 l/h per kilogram); the developed model successfully predicted this increase in CL after intravenous and oral administration of theophylline (Table 2) (Ellis et al., 1976; Loughnan et al., 1976; O’Hara, 2016). This increase in pediatric theophylline CL is associated with the age-related physiologic changes occurring in this population (from 1 to 10 years). The increase in pediatric CL is supported by the fact that children have a higher liver weight to body weight ratio than adults, which increases their capacity to clear administered drugs (Kanamori et al., 2002). Moreover, the major metabolic enzymes involved in the metabolism of theophylline (CYP1A2 and CYP3A4) reach adult equivalent values within the first few years after birth (Ginsberg et al., 2002; Salem et al., 2014).
To see the impact of HSA changes on theophylline PK in children with asthma, the unbound theophylline concentration was also predicted. There were no changes seen in the predicted unbound AUC0–last after intravenous and oral administration with and without incorporating HSA changes in children with asthma. The unbound AUC0-last in children with asthma after administering 3.2 and 4 mg/kg intravenous theophylline with and without changing HSA was 18.80 and 24.67 mg/l.h. Similarly, after administering 8.2 mg/kg oral theophylline, the unbound AUC0-last with and without changing HSA was 57.50 mg/l.h.
Here, it is worth mentioning that children are not young adults and that the developmental changes occurring in children can potentially affect the ADME of administered drugs ((Verscheijden et al., 2020)Verscheijden et al., 2020). Several factors can influence the disposition of administered drugs in the pediatric population. Some of these factors, such as the age-dependent changes in tissue composition ((Fernandez et al., 2011)Fernandez et al., 2011), the ontogeny of different metabolic enzymes, and the decreased protein binding, are already known ((Johnson et al., 2006)Johnson et al., 2006; (Abduljalil et al., 2014)Abduljalil et al., 2014; Salem et al., 2014; (Jones, 2018)Jones, 2018). However, the other disease-related factors that may affect the ADME of administered drugs in children that are not very well known are variability in organ blood flow, the abundance of various drug transporters, and changes in protein binding. The limited information on disease pathology and developmental biology in children are the two major challenges being faced during the development of pediatric PBPK models ((Cheung et al., 2019)Cheung et al., 2019; Verscheijden et al., 2020). Therefore, the predictive performance of the developed pediatric PBPK model should be assessed on the above-stated argument.
There are a few published reports of PBPK models for theophylline in children and adults (Ginsberg et al., 2004; Björkman, 2005). One of the published models was focused on predicting theophylline disposition in infants and children after administration of intravenous theophylline only (Björkman, 2005), and the other was used for risk assessment from environmental agents (Ginsberg et al., 2004). On the other hand, the presented work is focused on developing and evaluating the PBPK model for theophylline in adult (healthy and asthma) and pediatric (asthma) populations after incorporating changes in HSA. Moreover, by allowing the incorporation of in vitro dissolution profile of theophylline pellets in the ADAM model (Fig. 1N), the presented theophylline PBPK model can provide additional advantage in predicting ADME of novel theophylline dosage forms. Furthermore, by using a systematic model building approach, the presented model may help in understanding theophylline ADME after intravenous and oral application in different healthy and diseased populations.
The developed drug-disease PBPK model for theophylline has efficiently predicted theophylline PK in asthma adult and pediatric population. The incorporation of HSA changes in the asthma population did not result in the improvement of predictions. The Robs/Pre for all the PK parameters, i.e., AUC0-last, Cmax, CL, and CL/F, were within a 2-fold error range. Moreover, the AFE and RMSE values show that the developed model predicted theophylline PK accurately and precisely (Supplemental Table 4).
The mechanistic nature of the developed PBPK model can help in its extension to other drugs (high and low clearance) being used in the management of asthma. Moreover, it can also assist in the optimization of novel theophylline dosage forms.
Acknowledgments
The authors thank Certara UK Limited (Simcyp Division, Sheffield, UK) for granting free access to the Simcyp simulators through an academic license (subject to conditions).
Authorship Contributions
Participated in research design: Rasool, Khalid, Imran, Majeed, Saeed, Alasmari, Alanazi, Alqahtani.
Conducted simulations: Rasool, Khalid, Alqahtani.
Performed data analysis: Rasool, Khalid, Imran, Majeed, Saeed, Alasmari, Alanazi, Alqahtani.
Wrote or contributed to the writing of the manuscript: Rasool, Khalid, Imran, Majeed, Saeed, Alasmari, Alanazi, Alqahtani.
Footnotes
- Received February 18, 2020.
- Accepted April 20, 2020.
The authors extend their appreciation to the Deanship of Scientific Research at King Saud University for funding this work through Research Group No. RG-1441-340.
↵This article has supplemental material available at dmd.aspetjournals.org.
Abbreviations
- ADAM
- advanced; dissolution, absorption; and metabolism
- ADME
- absorption, distribution; metabolism, and eliminition
- AFE
- average fold error
- AUC0-last
- area under the systemic drug concentration-time curve from time zero to last measured systemic drug concentration
- CI
- confidence interval
- CL
- clearance
- CL/F
- clearance after oral application
- HSA
- human serum albumin
- P450
- cytochrome P450
- PBPK
- physiologically based pharmacokinetic
- PK
- pharmacokinetics
- RMSE
- root mean square error
- RObs/Pre
- ratio of observed and predicted
- VPC
- visual predictive check
- Copyright © 2020 by The American Society for Pharmacology and Experimental Therapeutics