- Email: [email protected]
Integration of In Silico Pharmacokinetic Modeling Approaches Into In Vitro Dissolution Profiles to Predict Bioavailability of a Poorly Soluble Compound
Journal of Pharmaceutical Sciences 108 (2019) 3723-3728
Contents lists available at ScienceDirect
Journal of Pharmaceutical Sciences journal homepage: www.jpharmsci.org
Pharmacokinetics, Pharmacodynamics and Drug Transport and Metabolism
Integration of In Silico Pharmacokinetic Modeling Approaches Into In Vitro Dissolution Profiles to Predict Bioavailability of a Poorly Soluble Compound Takafumi Kato 1, *, Tomoyuki Watanabe 2, Koichi Nakamura 3, Shuichi Ando 1 1 2 3
Formulation Technology Research Laboratories, Pharmaceutical Technology Division, Daiichi Sankyo Co., Ltd., Tokyo, Japan CMC Regulatory Affairs Department, Pharmaceutical Technology Division, Daiichi Sankyo Co., Ltd., Tokyo, Japan Drug Metabolism & Pharmacokinetics Research Laboratories, R&D Division, Daiichi Sankyo Co., Ltd., Tokyo, Japan
a r t i c l e i n f o
a b s t r a c t
Article history: Received 13 May 2019 Revised 23 June 2019 Accepted 25 June 2019 Available online 3 July 2019
The objective of present study is to develop pharmacokinetic (PK) prediction methods using in silico PK model for oral immediate release drug products (i.e. solution, suspension, and amorphous solid dispersion). A poorly water soluble compound with low bioavailability in rat was used (CS-758 as a model compound). A constructed in silico PK model contained an advance compartmental absorption and transit model. For solution, the in silico PK model reproduced an observed rat plasma concentration (Cp)-time profile. In addition, an in vitro dissolution method was developed to predict a rat Cp-time profile for suspension. As a result, the in silico PK model could predict the observed one by using dissolution profiles as the input. Furthermore, a dissolution profile of amorphous solid dispersion was applied to verify the in silico PK model. A result indicated the simulated rat Cp-time profile was significantly comparable to the observed one. This study demonstrated that the integration of an in silico PK model into dissolution profiles can predict rat Cp-time profiles for suspension and amorphous solid dispersion. These results suggest that the integration of in silico PK modeling approaches into dissolution profiles can contribute to the formulation screening for poorly soluble compounds by predicting PK behaviors. © 2019 American Pharmacists Association®. Published by Elsevier Inc. All rights reserved.
Keywords: poorly soluble compound pharmacokinetics in silico PK simulation advance compartmental absorption and transit model GastroPlusTM in vitro dissolution prediction accuracy
Introduction In pharmaceutical development, formulation designs for accomplishment of intended drug absorption are important to ensure the therapeutic effect, safety, and clinical success. To achieve the desirable drug efficacy and safety, general oral dosage forms, such as tablets or capsules, must disintegrate and dissolve in the intestinal fluids before crossing intestinal membrane. The dissolution of drug products is crucial for the pharmaceutical development, especially for a poorly water soluble compound, because this is an indicator to predict the in vivo drug release and absorption. In decades past, the development of in vitro dissolution approaches where reproduce in vivo dissolution behaviors have been attempted to ensure the appropriate drug bioavailability because understanding in vivo dissolution behaviors is one of the key roles to capture the drug disposition for orally administered solid dosage forms accurately.1-12
* Correspondence to: Takafumi Kato (Telephone: þ81 03 3492 3131). E-mail address: [email protected] (T. Kato).
Establishment of an in vitro/in vivo correlation (IVIVC) has been attempted as one of the approaches to predict in vivo bioavailability characteristics for orally administered solid dosage forms.13,14 However, development of the in vitro dissolution method considering absorption, distribution, metabolism, and excretion (ADME) behaviors is desirable as an advanced IVIVC approach to predict the drug bioavailability accurately. In addition, establishment of the advanced IVIVC using a paddle apparatus according to JP/USP/ Ph.Eur is challenging effort because the in vitro dissolution method is pharmacopeially used as a quality control tools to justify batchto-batch reproducibility of pharmaceutical solid dosage forms.15-17 In recent years, an integration of in silico pharmacokinetic (PK) modeling approaches into in vitro dissolution profiles has been studied as an advanced IVIVC approach. For example of an in silico PK model, an advance compartmental absorption and transit (ACAT) model is used to predict the gastrointestinal transit and absorption of compounds. The ACAT model is divided into 9 different compartments of the gut and can estimate the fraction absorption in each compartment. The ACAT model is considered as a prospective tool to simulate the PK behavior because not only
https://doi.org/10.1016/j.xphs.2019.06.026 0022-3549/© 2019 American Pharmacists Association®. Published by Elsevier Inc. All rights reserved.
T. Kato et al. / Journal of Pharmaceutical Sciences 108 (2019) 3723-3728
in vivo dissolution but also subsequent ADME behaviors in clinical situations could be predicted accurately.18 However, in current research and development, the most of in silico PK modeling approaches have been developed to predict the PK for the immediate release drug products or modified release drug products with water soluble compounds (e.g. Biopharmaceutics Classification System [BCS] class I or class III compounds).19,20 If these in silico approaches could also be applied for the drug products with poorly water soluble compounds (e.g. BCS class II or class IV compounds), they will be helpful approaches for the future formulation development such as formulation change and selection of dose strength. In addition, the justification of the integrated in silico PK model into in vitro dissolution profiles should be discussed. Furthermore, the prediction accuracy of the developed in silico PK model should be evaluated to assure the reliability of the model. The objective of present study is to develop reliable PK prediction method for oral immediate release drug products (i.e. solution, suspension, and amorphous solid dispersion). A poorly water soluble compound with low bioavailability in rat was used (CS-758 as a model compound). In this research, an in vitro dissolution condition was developed and verified by integration of in silico PK model into in vitro dissolution profiles. Materials and Methods Materials CS-758 was synthesized by Daiichi Sankyo Company, Ltd. (Tokyo, Japan). Methyl cellulose (MC) #400 and sodium lauryl sulfate (SLS) were purchased from Nacalai Tesque Inc. (Kyoto, Japan). Second fluid for disintegration test (JP2) as phosphate buffer solution was purchased from Kanto Chemical Company, Inc. (Tokyo, Japan). Polyethyleneglycol (PEG) 400 and polysorbate 80 were purchased from Sigma-Aldrich (St. Louis, MO). Polyvinylpyrrolidone (PVP) K-30 was purchased from BASF Japan Ltd. (Tokyo, Japan). Methanol were all of JIS special grade and were purchased from Wako Pure Chemical Industries, Ltd. (Osaka, Japan). Methods Preparation of CS-758 Solution and Suspension CS-758 solution was prepared by completely dissolving 10 mg of CS-758 drug substance with PEG 400/polysorbate 80 (1:1 [w/w]) solvent. CS-758 suspension was prepared by homogenously suspending 10 mg of CS-758 drug substance with 10 mL of 0.5% MC. Preparation of CS-758 Amorphous Solid Dispersion by Spray Drying The solubility of CS-758 is extremely low, and the rat bioavailability was approximately 53%.21 The amorphous solid dispersion has been developed as practical method to overcome the bioavailability of poorly water soluble drugs.22 The amorphous solid dispersion is the one of the active areas of research in the pharmaceutical field23 and is potential to enhance the bioavailability of CS-758. CS-758 amorphous solid dispersion was prepared by spray drying using CS-758 drug substance and a polymer to improve both the solubility and bioavailability. A physical mixture was prepared by mixing CS-758 with PVP K30 in a weight ratio of 1:1 using a mortar and a pestle. The physical mixture was dissolved in methanol. An amorphous solid dispersion composed of CS-758:PVP K-30 (1:1 (w/w)) was prepared using a Mini Spray Dryer B-290 (BÜCHI Labortechnik AG, Flawil, Switzerland). The inlet air temperature of 100 C, the exhaust air temperature of 60 C, the aspirator of 100%, and the spray rate of 15 mL/min were set as the spray drying conditions using B-290.
Powder X-ray Diffraction The powder X-ray diffraction (PXRD) patterns of the samples were measured using a Geiger Flex Rint-2200 diffractometer (Rigaku Corporation, Tokyo, Japan) with the CuKa radiation at 40 kV/40 mA. The samples were step-scanned at 0.02 intervals from 5.00 to 40.00 (2q) at the rate of 4.00 min1. The polymorph of CS-758 drug substance in the amorphous solid dispersion was characterized using PXRD. The PXRD patterns of unprocessed CS-758 drug substance, physical mixture blended with the drug substance and PVP K-30, and the amorphous solid dispersion (CS-758:PVP K-30 ¼ 1:1 [w/w]) are illustrated (Fig. S1). The unprocessed drug substance and the physical mixture showed sharp crystalline peaks. In contrast, the amorphous solid dispersion showed an amorphous halo pattern only. These results indicated that the polymeric structure of CS-758 drug substance in amorphous solid dispersion was amorphous. In Vitro Dissolution USP Apparatus 2 (Paddle Apparatus) described in USP <711> with UV-VIS analysis was used for in vitro dissolution testing. Each dissolution vessel was filled with 500 mL of dissolution medium of pH 6.9 phosphate buffer or pH 6.9 phosphate buffer with various concentration of SLS (0.1%, 0.15%, and 0.2% w/v), and the medium in the vessel was kept at 37 C. The stirring speed of paddle was set at 75 rpm. The dissolution testing (n ¼ 3) was conducted using 125 mg of CS-758 drug substance for all dosage forms. For generating the dissolution profiles, the dissolution test sample was taken from the vessel and filtered with a membrane filter (0.45 mm pore size) at 5, 10, 15, 20, 30, 45, 60, 90, 120, 150, 180, 210, 240, 270, 300, 330, and 360 min. The filtered sample was carried directly to the flow cell of the circulation system. The sample concentration was determined by UV-VIS spectrophotometry, and then the samples was returned to the vessel. In Vivo Animal Pharmacokinetic Experiments All animal experiments were performed in accordance with the in-house guidelines provided by the Institutional Animal Care and Use Committee of Daiichi Sankyo Company, Ltd. Male rats (n ¼ 3) were dosed in a nonrandomized, single-sequence design and fasted overnight. In a preparation of solution, 10 mg of CS-758 was completely dissolved with 4 mL of PEG 400/polysorbate 80 (1:1 [w/w]) solvent, and the saline was added up to 10 mL. In preparations of the suspension and the amorphous solid dispersion, 10 mg
1000
Plasma concentration (ng/mL)
3724
900
Observed
800
Simulated
700 600 500 400 300 200 100 0 0
3
6
9
12 15 Time (hr)
18
21
24
Figure 1. Observed and simulated rat plasma concentrationetime profiles after orally administered with CS-758 solution. The simulated profile (solid line) was obtained using ACAT model based on the physicochemical and pharmacokinetic properties, and the observed rat plasma concentrationetime profiles (closed diamond) after orally administered with CS-758 solution.
T. Kato et al. / Journal of Pharmaceutical Sciences 108 (2019) 3723-3728
3725
Table 1 Cmax, AUC0-t, and Tmax Predicted From a Simulated Rat Cp-Time Profile and Those Calculated From an Observed Rat Cp-Time Profile After Orally Administered With CS-758 Solution. %PEs of Cmax and AUC0-t Were Calculated Based on the Observed and Simulated Data Parameters
Observeda
Simulated
Absolute Percent Prediction Error (%PE)
Cmax (ng/mL) AUC0-t (ng/mL$hr) Tmax (hr)
703 7963 2
744 7907 2
6 1 -
%PE was evaluated as a reference of Guidance28 for Industry, Extended Release Oral Dosage Forms: Development, Evaluation, and Application of In Vitro/In Vivo Correlations. a Calculated by GastroPlus™.
of CS-758 or 20 mg of amorphous solid dispersion were suspended with 10 mL of 0.5% MC. Rat pharmacokinetic experiments were conducted under fasting conditions because bioavailability and bioequivalence studies for immediate release products are generally conducted under this condition. CS-758 solution or suspension were orally administered at a dose of 5 mg/5 mL/kg under fasting conditions. CS-758 amorphous solid dispersion was orally administered at a dose of 10 mg/5 mL/kg under fasting conditions. The dosing for these dosage forms was administered in a single oral dose via a catheter connected to a syringe into the stomach. After that, the syringe was exchanged to a new one filled with tap water and then the water was given to the rats. Blood samples were collected from jugular vein with a heparinized syringe at before administration (pre) or 0.5, 1, 2, 4, 8, and 24 h after administration. The collected blood was transferred to a microtube and put on ice immediately. The blood samples were centrifuged at 15,000 rpm for 5 min to separate the plasma. The supernatant was collected and stored at -20 C until analysis. The concentration of CS-758 in the supernatant was measured by liquid chromatography coupled with tandem mass spectrometry with the high-performance liquid chromatography system. In Silico Pharmacokinetic Modeling The rat Cp-time profiles for CS-758 were simulated using the commercially available software, GastroPlus™, version 9.0 (Simulations Plus Inc., Lancaster, CA). An ACAT model considering in vivo dissolution behaviors and gastrointestinal transit can be constructed using this software.24 An in silico PK model composed of an ACAT model was constructed using the physicochemical properties of CS-758 and the estimated PK parameter values. Input parameters used to construct the in silico PK model were set based on the in vitro experimental results and the values predicted by the ADMET predictor™. ADMET
predictor™ is commercially available software package to predict physicochemical and pharmacokinetic parameters based on structure formula of molecule.25 The input parameters used to construct the in silico PK model are summarized (Table S1). In the in silico PK model, the “IR: Solution” as dosage form option in Compound tab, and the “Rat-Physiological-Fasted” as Physiology and the “Opt logD Model SA/V 6.1” as ASF model in the Gut-physiology tab were selected. The input parameters of Log P, pKa, Peff (permeability parameter), blood/Cp ratio, and fraction unbound in plasma were predicted and set by ADMET predictor™. Clearance, volume of distribution, and the 2-compartment rate constants, k12 and k21, were calculated by fitting the simulated results to the observed rat Cp-time profile after orally administered with CS-758 solution using optimization module of GastroPlus™. Remaining model parameters used for this model were obtained as default setting parameters as shown (Table S1). For the integration of the in silico PK model into in vitro dissolution profiles, “CR: Dispersed” as dosage form option in Compound tab was selected. This option assumes that the unreleased drug is remained in gastrointestinal tract instead of disintegrating into individual particles and dispersing through the entire gastrointestinal tract. The simulation for rat Cp-time profiles using in vitro dissolution profiles were conducted using Weibull function in GastroPlus™. The release rate of the drug substance using Weibull function was calculated by the following equation.26,27
b A %Dose Released ¼ Max 1 f exp t Tlag where Max is percentage of maximum dissolution release, Tlag is lag time for dissolution, f is fraction of dissolution profile, t is time, A is the time scale factor, and b is the shape factor.
pH 6.9 phosphate buffer pH 6.9 phosphate buffer with 0.1%SLS pH 6.9 phosphate buffer with 0.15%SLS pH 6.9 phosphate buffer with 0.2%SLS
Plasma concentration (ng/mL)
Dissolution (%)
250 100 90 80 70 60 50 40 30 20 10 0
Observed Simulated(pH 6.9 phosphate buffer) Simulated(pH 6.9 phosphate buffer with 0.1%SLS) Simulated(pH 6.9 phosphate buffer with 0.15%SLS) Simulated(pH 6.9 phosphate buffer with 0.2%SLS)
200 150 100 50 0 0
0
30 60 90 120 150 180 210 240 270 300 330 360 Time (min)
Figure 2. Mean in vitro dissolution profiles (n ¼ 3) for CS-758 suspension using USP apparatus II in pH 6.9 phosphate buffer with various concentration of SLS (0%, 0.1%, 0.15%, and 0.2%). 500 mL of dissolution media was filled in vessels, and the media in the vessel were kept at 37 C. Stirring speed of paddle was set at 75 rpm. The dissolution testing was conducted using 125 mg of CS-758 drug substance for all dosage forms.
3
6
9
12 15 Time (hr)
18
21
24
Figure 3. Simulated rat plasma concentrationetime profiles after orally administered with CS-758 suspension using Weibull parameters fitted to the dissolution profiles in pH 6.9 phosphate buffer with various concentration of SLS (0%, 0.1%, 0.15%, and 0.2%). The rat plasma concentrationetime profiles of CS-758 suspension were predicted using the integrated in silico PK model into evaluated in vitro dissolution profiles as shown in Figure 2. The rat plasma concentration-time profiles of CS-758 suspension are shown as simulated profiles (solid lines) and an observed profile (closed diamond).
3726
T. Kato et al. / Journal of Pharmaceutical Sciences 108 (2019) 3723-3728
Evaluation of Prediction Accuracy for In Silico PK Model The absolute percent prediction error (%PE) was calculated by the following equation as the evaluation index for prediction accuracy of the established in silico PK model. The definition of %PE was referred from Guidance28 for Industry, Extended Release Oral Dosage Forms: Development, Evaluation, and Application of IVIVCs.
%PE ¼ ½ðObserved value Predicted valueÞ= Observed value 100 Based on the Guidance for Industry, the acceptance criteria of maximum plasma concentration (Cmax) and area under the plasma concentration-time curve (AUC0-t) predicted by the developed in silico PK model was established as 15% or less for %PE. Results and Discussion In Silico Pharmacokinetic Model Construction for CS-758 Solution The observed and simulated rat Cp-time profiles after orally administered with CS-758 solution were shown (Fig. 1). The Cmax, AUC0-t, and time to maximum plasma concentration (Tmax) calculated from observed and simulated Cp-time profiles were summarized (Table 1). As a result, the %PEs was calculated for Cmax (6%) and for AUC0-t (1%). The results indicated that the simulated Cmax and AUC0-t were reliable prediction accuracy because the %PEs for Cmax and AUC0-t met the predefined acceptance criterion, less than 15%. Therefore, the in silico PK model was judged to have a reliable prediction accuracy to predict the rat Cp-time profiles for CS-758 solution. Based on the result, the in silico PK model was considered to be applied for developing the in vitro dissolution method. Development of In Vitro Dissolution Method to Predict the Rat Plasma Concentration-Time Profiles of CS-758 Suspension To develop the in vitro dissolution method, the condition was explored to predict a rat Cp-time profile for CS-758 suspension. The mean dissolution profiles for CS-758 suspension using USP apparatus 2 in pH 6.9 phosphate buffer with various concentration of SLS (0%, 0.1%, 0.15%, and 0.2%) were shown (Fig. 2). The SLS is generally used as a surfactant to accomplish the desired dissolution profiles for poorly water-soluble compounds. The SLS was added to the dissolution medium for the in vitro dissolution method development because the solubility of CS-758 was extremely low (0.05 mg/mL in phosphate buffer) as shown (Table S1). The results indicated that an in vitro dissolution profile of CS-758 suspension in pH 6.9 phosphate buffer without SLS was extremely low (approximately 1% of dissolution rate at 360 min). The dissolution rate was increased with the addition of SLS, the more SLS concentrations the higher dissolution
rate. The dissolution rate at 360 min was approximately 23% when the SLS concentration in the dissolution medium was 0.2%. Based on the in vitro dissolution results, the obtained in vitro dissolution profiles were incorporated into the constructed in silico PK model to predict a rat Cp-time profile for CS-758 suspension. Weibull parameters were determined by fitting to the mean in vitro dissolution profiles using GastroPlus™ software. The determined Weibull parameters with various concentration of SLS in pH 6.9 phosphate buffer are summarized (Table S2). These determined Weibull parameters were used as input parameters for the in silico PK simulation. Then, the in vitro dissolution condition with the best prediction accuracy was selected based on the result of %PE evaluation. The observed and simulated rat mean Cp-time profiles after orally administered with CS-758 suspension are shown (Fig. 3). The %PEs for the Cmax and AUC0-t for CS-758 suspension evaluated using GastroPlus™ were summarized (Table 2). The %PEs were calculated for Cmax (4%) and AUC0-t (0.5%) when the in vitro dissolution profile in pH 6.9 phosphate buffer with 0.15% SLS concentration was incorporated into the constructed in silico PK model. On the other hand, the %PEs for the Cmax and AUC0-t were not reliable prediction accuracy (more than 30% of %PEs) when the in vitro dissolution profiles in pH 6.9 phosphate buffer with the other SLS concentration were incorporated into the constructed in silico PK model. The results indicated that %PEs for the Cmax and AUC0-t calculated using the dissolution profile of pH 6.9 phosphate buffer with 0.15% SLS met the predefined acceptance criterion, less than 15%. Based on the results, USP Apparatus 2 (Paddle Apparatus) of paddle rotation speed of 75 rpm and 500 mL of pH 6.9 phosphate buffer with 0.15% SLS was applied as a dissolution method that can predict the rat Cp-time profile of CS-758 suspension. The dissolution of CS-758 drug substance in the suspension with this in vitro dissolution condition did not show the complete dissolution because the dissolution rate at 360 min was low (approximately 16%). However, the integration of in silico PK model into this dissolution profile had the desirable prediction accuracy to reflect the rat Cp-time profile of CS-758 suspension. Therefore, the in vivo dissolution behavior of the CS-758 suspension in the intestine was also considered to indicate an incomplete dissolution behavior because CS-758 drug substance was of low solubility. Moreover, in this investigation, an addition of SLS in the dissolution medium was considered to be optimal to predict the rat Cp-time profile of CS-758 suspension although the SLS is not included as the components in gastric and intestinal fluids. The predictable in silico PK model could be developed by an integration of the in silico model into in vitro dissolution profiles of CS-758 suspension because this model was confirmed to have the desirable prediction accuracy. These results suggest that the integration of in silico PK model into in vitro dissolution profiles can be applied to predict the rat Cp-time profiles for the other formulation (i.e. amorphous solid dispersion), with different in vitro dissolution behaviors.
Table 2 Cmax, AUC0-t, and Tmax Predicted From Simulated Rat Cp-Time Profiles Using In Vitro Dissolution Data in pH 6.9 Phosphate Buffer With Various Concentration of SLS and Those Calculated From an Observed Cp-Time Profile After Orally Administered With CS-758 Suspension. %PEs of Cmax and AUC0-t Were Calculated Based on the Observed and Simulated Data Parameters
Cmax (ng/mL) AUC0-t (ng/ mL$hr) Tmax (hr)
Observeda Without SLS
0.1% SLS
0.15% SLS
0.2% SLS
Simulated Absolute Percent Prediction Error (%PE)
Simulated Absolute Percent Prediction Error (%PE)
Simulated Absolute Percent Prediction Error (%PE)
Simulated Absolute Percent Prediction Error (%PE)
103
7.9
92
46
55
99
4
136
32
1315
120
91
638
51
1309
0.5
1903
45
4
4
-
4
-
4
-
4
-
%PE was evaluated as a reference of Guidance a Calculated by GastroPlus™.
28
for Industry, Extended Release Oral Dosage Forms: Development, Evaluation, and Application of In Vitro/In Vivo correlations.
Dissolution (%)
T. Kato et al. / Journal of Pharmaceutical Sciences 108 (2019) 3723-3728
100 90 80 70 60 50 40 30 20 10 0
CS-758 suspension CS-758 amorphous solid dispersion (CS-758:PVP K-30 =1:1 (w/w))
0
30 60 90 120 150 180 210 240 270 300 330 360 Time (min)
Figure 4. Dissolution profiles (n ¼ 3) of CS-758 suspension (cross) and amorphous solid dispersion (closed circle) in pH 6.9 phosphate buffer with 0.15% SLS. 500 mL of dissolution medium was filled in vessels and the medium in the vessel was kept at 37 C. Stirring speed of paddle was set at 75 rpm. The dissolution testing was conducted using 125 mg of CS-758 drug substance for all dosage forms.
Verification of a Developed In Silico PK Model by Predicting Rat Plasma Concentration-Time Profiles of CS-758 Amorphous Solid Dispersion In this investigation, the developed in silico PK modeling approaches that were integrated into in vitro dissolution method were further evaluated as a verification process by predicting a rat Cptime profile for CS-758 amorphous solid dispersion. The in vitro dissolution profiles of the CS-758 suspension and amorphous solid dispersion in pH 6.9 phosphate buffer with 0.15% SLS concentration were shown (Fig. 4). The result showed that the in vitro dissolution profile of CS-758 amorphous solid dispersion was higher than that of CS-758 suspension. The in vitro dissolution rate of CS-758 was considered to be increased by an effect of the amorphous state of CS-758 drug substance in the amorphous solid dispersion. Based on the in vitro dissolution results, the obtained in vitro dissolution profiles were incorporated into the constructed in silico PK model to predict a rat Cp-time profile for CS-758 amorphous solid dispersion. Weibull parameters were determined by fitting to a mean in vitro dissolution profile using GastroPlus™ software. The determined Weibull parameters are summarized (Table S3). The observed and simulated rat mean Cp-time profiles of CS-758 suspension and amorphous solid dispersion are illustrated (Fig. 5). The simulated Cmax and AUC0-t for CS-758 amorphous solid dispersion were higher than those for the CS-758 suspension when the in vitro dissolution profile for CS-758 amorphous solid dispersion was
incorporated into the constructed in silico PK model (approx. 3 times higher) (Table 2 vs. Table 3). The results indicated that the bioavailability of the amorphous solid dispersion would be predicted to be improved in comparison with that of the CS-758 suspension. The %PEs for the Cmax and AUC0-t for CS-758 amorphous solid dispersion evaluated using GastroPlus™ were summarized (Table 3). A result indicated that the simulated Cp-time profile for the amorphous solid dispersion was comparable to the observed one because the %PEs for the Cmax (9%) and AUC0-t (11%) met the predefined acceptance criterion, less than 15%. Therefore, it was demonstrated that the in silico PK model integrated into in vitro dissolution profiles had a reliable prediction accuracy for rat Cptime profiles for suspension and amorphous solid dispersion. The predictable in silico PK model could be verified by an integration of the model into an in vitro dissolution profile of CS-758 amorphous solid dispersion because this in silico model was also confirmed to have the desirable prediction accuracy. This in silico modeling approaches were judged to be predictable approaches to predict in vivo PK behaviors by development and verification processes. It has been studied that an IVIVC approach using an in vitro dissolution profiles and in silico PK model could predict PK behaviors of modified release formulation for BCS class I compound.19 Moreover, it was reported that a mechanistic-based IVIVC approach using an in vitro dissolution profiles and in silico PK model for BCS class III compound was developed for demonstrating acceptable Cp-time profiles. A strategy for considering biowaiver approach was provided by this mechanistic-based approach.20 These reports concluded that IVIVC approaches using an integration of in silico PK model into in vitro dissolution profiles were applicable to simulate Cp-time profiles for water soluble compounds (e.g. BCS class I and III). Based on these reports, the investigational results in this section supported that the integration of in silico PK model into in vitro dissolution profiles can also be useful for formulation screening considering PK behaviors for poorly soluble compounds (e.g. BCS class II or IV). Therefore, the integration of in silico PK modeling approaches will be considered to be applicable for the other poorly soluble compounds, depending on the physicochemical properties and PK behaviors. The conventional IVIVC relationship should be demonstrated consistently with 2 or more formulations with different release rates to result in corresponding differences in absorption profiles. Although an IVIVC can be defined with a minimum of 2 formulations with different release rates, 3 or more formulations with different release rate are recommended.28 On the other hand, the integration of in silico PK modeling approaches into in vitro dissolution profiles could be constructed based on one dosage form with one in vitro dissolution profile. Therefore, the in silico PK modeling
500
Plasma concentration (ng/mL)
3727
Observed (CS-758 suspension)
450 400 350
Simulated (CS-758 suspension)
300 250
Observed (CS-758 amorphos solid dispersion (CS-758:PVP K-30=1:1 (w/w))
200 150
Simulated (CS-758 amorphos solid dispersion (CS-758:PVP K-30=1:1 (w/w))
100 50 0 0
3
6
9
12 15 Time (hr)
18
21
24
Figure 5. Simulated (plots) and observed (solid lines) rat plasma concentrationetime profiles after orally administered with CS-758 suspension and amorphous solid dispersion. The rat plasma concentrationetime profiles of CS-758 suspension and amorphous solid dispersion were predicted using the integrated in silico PK model into evaluated in vitro dissolution profiles as shown in Figure 4. The rat plasma concentrationetime profiles of CS-758 suspension and amorphous solid dispersion are shown as simulated profiles (blue and orange solid lines) and an observed profile (closed diamond and cross).
3728
T. Kato et al. / Journal of Pharmaceutical Sciences 108 (2019) 3723-3728
Table 3 Cmax, AUC0-t, and Tmax Predicted From a Simulated Rat Cp-Time Profile and Those Calculated From an Observed Rat Cp-Time Profile After Orally Administered With CS-758 Amorphous Solid Dispersion. %PEs of Cmax and AUC0-t Were Calculated Based on the Observed and Simulated Data Parameters
Observeda
Simulated
Absolute Percent Prediction Error (%PE)
Cmax (ng/mL) AUC0-t (ng/mL$hr) Tmax (hr)
299 3893 4
267b 4239c 5
11 9 -
%PE was evaluated as a reference of Guidance28 for Industry, Extended Release Oral Dosage Forms: Development, Evaluation, and Application of In Vitro/In Vivo Correlations. a Calculated by GastroPlus™. b 2.7 times for simulated Cmax for CS-758 suspension. c 3.2 times for simulated AUC0-t for CS-758 suspension.
approaches can construct with less in vitro dissolution profile data compared with the conventional IVIVC approaches. The in silico PK modeling approaches can be useful for finding optimal formulation to accomplish target product profiles (target dissolution profiles) and target PK behaviors with the desirable prediction accuracy. Conclusions The investigation of stepwise integration of in silico PK modeling approaches into in vitro dissolution profiles was demonstrated to reproduce the observed PK profiles for oral immediate release drug products (i.e. solution, suspension, and amorphous solid dispersion). The in silico PK model was constructed using observed PK data for CS-758 solution. In addition, an in vitro dissolution method was developed to predict a rat Cp-time profile for suspension. Furthermore, the PK profile of CS-758 amorphous solid dispersion was accurately predicted based on the %PE metrics. These approaches can contribute to the formulation screening for poorly soluble compounds (e.g. BCS class II or IV) by predicting PK behaviors with the dissolution profile, even if the in vitro dissolution methods are pharmacopeially applied as a quality control tools. Same approaches can be applied and useful for human PK behaviors using observed human Cp-time profiles. Acknowledgments € hling The authors would like to acknowledge Pearnchob-Ho Nantharat, Dr. and Hiroshi Nakagawa, Dr. in Daiichi Sankyo Europe GmbH, and Tsuyoshi Mikkaichi in Daiichi Sankyo Co., Ltd. for valuable consultations regarding preparation for this scientific article. The authors would like to acknowledge Yumiko Fujii for valuable consultations regarding submitting this scientific article. References 1. Dressman JB, Reppas C. In vitroein vivo correlations for lipophilic, poorly water-soluble drugs. Eur J Pharm Sci. 2000;11:S73-S80. 2. Balan G, Timmins P, Greene DS, Marathe PH. In-vitro in-vivo correlation models for glibenclamide after administration of metformin/glibenclamide tablets to healthy human volunteers. J Pharm Pharmacol. 2000;52:831-838. 3. Nicolaides E, Symillides M, Dressman JB, Reppas C. Biorelevant dissolution testing to predict the plasma profile of lipophilic drugs after oral administration. Pharm Res. 2001;18:380-388.
4. Balan G, Timmins P, Greene DS, Marathe PH. In vitroein vivo correlation (IVIVC) models for metformin after administration of modified-release (MR) oral dosage forms to healthy human volunteers. J Pharm Sci. 2001;90:1176-1185. 5. Sirisuth N, Augsburger LL, Eddington ND. Development and validation of a nonlinear IVIVC model for a diltiazem extended release formulation. Biopharm Drug Dispos. 2002;23:1-8. 6. Sunesen VH, Pedersen BL, Kristensen HG, Mullertz A. In vivo in vitro correlations for a poorly soluble drug, danazol, using the flow-through dissolution method with biorelevant dissolution media. Eur J Pharm Sci. 2005;24:305-313. 7. Ghosh A, Bhaumik UK, Bose A, et al. Extended release dosage form of glipizide: development and validation of a level A in vitroein vivo correlation. Biol Pharm Bull. 2008;31:1946-1951. 8. Macha S, Yong CL, Darrington T, et al. In vitroein vivo correlation for nevirapine extended release tablets. Biopharm Drug Dispos. 2009;30:542-550. 9. Rossi RC, Dias CL, Bajerski L, et al. Development and validation of discriminating method of dissolution for fosamprenavir tablets based on in vivo data. J Pharm Biomed Anal. 2011;54:439-444. 10. Saibi Y, Sato H, Tachiki H. Developing in vitroein vivo correlation of risperidone immediate release tablet. AAPS PharmSciTech. 2012;13:890-895. 11. Guhmann M, Thommes M, Gerber F, et al. Design of biorelevant test setups for the prediction of diclofenac in vivo features after oral administration. Pharm Res. 2013;30:1483-1501. 12. Kesisoglou F, Rossenu S, Farrell C, et al. Development of in vitro-in vivo correlation for extended-release niacin after administration of hypromellosebased matrix formulations to healthy volunteers. J Pharm Sci. 2014;103: 3713-3723. 13. Li S, He H, Parthiban LJ, et al. IV-IVC considerations in the development of immediate-release oral dosage form. J Pharm Sci. 2005;94:1396-1417. 14. Bredael GM, Bowers N, Boulineau F, et al. In vitroein vivo correlation strategy applied to an immediate-release solid oral dosage form with a Biopharmaceutical Classification System IV compound case study. J Pharm Sci. 2014;103:2125-2130. 15. Lagece M, Gravelle L, Di Maso M, et al. Developing a discriminating dissolution procedure for a dual active pharmaceutical product with unique solubility characteristics. Dissolution Technol. 2004;11:13-17. 16. He Z, Zhong D, Chen X, et al. Development of a dissolution medium for nimodipine tablets based on bioavailability evaluation. Eur J Pharm Sci. 2004;21:487-491. 17. Wu C, Liu Y, He Z, et al. Insight into the development of dissolution media for BCS Class II drugs: a review from quality control and prediction of in vivo performance perspectives. Curr Drug Deliv. 2016;13:1004-1020. 18. Kostewicz ES, Aarons L, Bergstrand M, et al. PK models for the prediction of in vivo performance of oral dosage forms. Eur J Pharm Sci. 2014;57:300-321. 19. Mirza T, Bykadi SA, Ellison CD, et al. Use of in vitroein vivo correlation to predict the pharmacokinetics of several products containing a BCS class 1 drug in extended release matrices. Pharm Res. 2013;30:179-190. 20. Crison JR, Timmins P, Keung A, et al. Biowaiver approach for biopharmaceutics Classification system class 3 compound metformin hydrochloride using in silico modeling. J Pharm Sci. 2012;101:1773-1782. 21. Kagoshima Y, Mori M, Suzuki E, et al. Synthesis, cleavage, and antifungal activity of a number of novel, water-soluble ester prodrugs of antifungal triazole CS-758. Biroog Med Chem. 2009;19:3559-3563. 22. Sekiguchi K, Obi N. Studies on absorption of eutectic mixture. I. A comparison of the behavior of eutectic mixture of sulfathiazole and that of ordinary sulfathiazole in man. Chem Pharm Bull. 1961;9:866-872. 23. Serajuddin ATM. Solid dispersion of poorly water-soluble drugs: early promises, subsequent problems, and recent breakthroughs. J Pharm Sci. 1999;88: 1058-1066. 24. Agoram B, Woltosz WS, Bolger MB. Predicting the impact of pharmacokinetic and biochemical processes on oral drug bioavailability. Adv Drug Deliv Rev. 2001;50:S41-S67. 25. Emoto C, Murayama N, Rostami-Hodjegan A, et al. Utilization of estimated physicochemical properties as an integrated part of predicting hepatic clearance in the early drug-discovery stage: impact of plasma and microsomal binding. Xenobiotica. 2009;39:227-235. 26. Wang B, Liu Z, Li D, et al. Application of physiologically based pharmacokinetic modeling in the prediction of pharmacokinetics of bicyclol controlled-release formulation in human. Eur J Pharm Sci. 2015;77:265-272. 27. Babiskin AH, Zhang X. Application of pharmacokineticly based absorption modeling for amphetamine salts drug products in generic drug evaluation. J Pharm Sci. 2015;104:3170-3182. 28. FDA. Guidance for industry. Extended Release Oral Dosage Forms: Development, Evaluation and Application of in Vitro/in Vivo Correlations. Rockville (MD): Center for Drug Evaluation and Research (CDER), US Department of Health and Human Services; 1997.
Contents lists available at ScienceDirect
Journal of Pharmaceutical Sciences journal homepage: www.jpharmsci.org
Pharmacokinetics, Pharmacodynamics and Drug Transport and Metabolism
Integration of In Silico Pharmacokinetic Modeling Approaches Into In Vitro Dissolution Profiles to Predict Bioavailability of a Poorly Soluble Compound Takafumi Kato 1, *, Tomoyuki Watanabe 2, Koichi Nakamura 3, Shuichi Ando 1 1 2 3
Formulation Technology Research Laboratories, Pharmaceutical Technology Division, Daiichi Sankyo Co., Ltd., Tokyo, Japan CMC Regulatory Affairs Department, Pharmaceutical Technology Division, Daiichi Sankyo Co., Ltd., Tokyo, Japan Drug Metabolism & Pharmacokinetics Research Laboratories, R&D Division, Daiichi Sankyo Co., Ltd., Tokyo, Japan
a r t i c l e i n f o
a b s t r a c t
Article history: Received 13 May 2019 Revised 23 June 2019 Accepted 25 June 2019 Available online 3 July 2019
The objective of present study is to develop pharmacokinetic (PK) prediction methods using in silico PK model for oral immediate release drug products (i.e. solution, suspension, and amorphous solid dispersion). A poorly water soluble compound with low bioavailability in rat was used (CS-758 as a model compound). A constructed in silico PK model contained an advance compartmental absorption and transit model. For solution, the in silico PK model reproduced an observed rat plasma concentration (Cp)-time profile. In addition, an in vitro dissolution method was developed to predict a rat Cp-time profile for suspension. As a result, the in silico PK model could predict the observed one by using dissolution profiles as the input. Furthermore, a dissolution profile of amorphous solid dispersion was applied to verify the in silico PK model. A result indicated the simulated rat Cp-time profile was significantly comparable to the observed one. This study demonstrated that the integration of an in silico PK model into dissolution profiles can predict rat Cp-time profiles for suspension and amorphous solid dispersion. These results suggest that the integration of in silico PK modeling approaches into dissolution profiles can contribute to the formulation screening for poorly soluble compounds by predicting PK behaviors. © 2019 American Pharmacists Association®. Published by Elsevier Inc. All rights reserved.
Keywords: poorly soluble compound pharmacokinetics in silico PK simulation advance compartmental absorption and transit model GastroPlusTM in vitro dissolution prediction accuracy
Introduction In pharmaceutical development, formulation designs for accomplishment of intended drug absorption are important to ensure the therapeutic effect, safety, and clinical success. To achieve the desirable drug efficacy and safety, general oral dosage forms, such as tablets or capsules, must disintegrate and dissolve in the intestinal fluids before crossing intestinal membrane. The dissolution of drug products is crucial for the pharmaceutical development, especially for a poorly water soluble compound, because this is an indicator to predict the in vivo drug release and absorption. In decades past, the development of in vitro dissolution approaches where reproduce in vivo dissolution behaviors have been attempted to ensure the appropriate drug bioavailability because understanding in vivo dissolution behaviors is one of the key roles to capture the drug disposition for orally administered solid dosage forms accurately.1-12
* Correspondence to: Takafumi Kato (Telephone: þ81 03 3492 3131). E-mail address: [email protected] (T. Kato).
Establishment of an in vitro/in vivo correlation (IVIVC) has been attempted as one of the approaches to predict in vivo bioavailability characteristics for orally administered solid dosage forms.13,14 However, development of the in vitro dissolution method considering absorption, distribution, metabolism, and excretion (ADME) behaviors is desirable as an advanced IVIVC approach to predict the drug bioavailability accurately. In addition, establishment of the advanced IVIVC using a paddle apparatus according to JP/USP/ Ph.Eur is challenging effort because the in vitro dissolution method is pharmacopeially used as a quality control tools to justify batchto-batch reproducibility of pharmaceutical solid dosage forms.15-17 In recent years, an integration of in silico pharmacokinetic (PK) modeling approaches into in vitro dissolution profiles has been studied as an advanced IVIVC approach. For example of an in silico PK model, an advance compartmental absorption and transit (ACAT) model is used to predict the gastrointestinal transit and absorption of compounds. The ACAT model is divided into 9 different compartments of the gut and can estimate the fraction absorption in each compartment. The ACAT model is considered as a prospective tool to simulate the PK behavior because not only
https://doi.org/10.1016/j.xphs.2019.06.026 0022-3549/© 2019 American Pharmacists Association®. Published by Elsevier Inc. All rights reserved.
T. Kato et al. / Journal of Pharmaceutical Sciences 108 (2019) 3723-3728
in vivo dissolution but also subsequent ADME behaviors in clinical situations could be predicted accurately.18 However, in current research and development, the most of in silico PK modeling approaches have been developed to predict the PK for the immediate release drug products or modified release drug products with water soluble compounds (e.g. Biopharmaceutics Classification System [BCS] class I or class III compounds).19,20 If these in silico approaches could also be applied for the drug products with poorly water soluble compounds (e.g. BCS class II or class IV compounds), they will be helpful approaches for the future formulation development such as formulation change and selection of dose strength. In addition, the justification of the integrated in silico PK model into in vitro dissolution profiles should be discussed. Furthermore, the prediction accuracy of the developed in silico PK model should be evaluated to assure the reliability of the model. The objective of present study is to develop reliable PK prediction method for oral immediate release drug products (i.e. solution, suspension, and amorphous solid dispersion). A poorly water soluble compound with low bioavailability in rat was used (CS-758 as a model compound). In this research, an in vitro dissolution condition was developed and verified by integration of in silico PK model into in vitro dissolution profiles. Materials and Methods Materials CS-758 was synthesized by Daiichi Sankyo Company, Ltd. (Tokyo, Japan). Methyl cellulose (MC) #400 and sodium lauryl sulfate (SLS) were purchased from Nacalai Tesque Inc. (Kyoto, Japan). Second fluid for disintegration test (JP2) as phosphate buffer solution was purchased from Kanto Chemical Company, Inc. (Tokyo, Japan). Polyethyleneglycol (PEG) 400 and polysorbate 80 were purchased from Sigma-Aldrich (St. Louis, MO). Polyvinylpyrrolidone (PVP) K-30 was purchased from BASF Japan Ltd. (Tokyo, Japan). Methanol were all of JIS special grade and were purchased from Wako Pure Chemical Industries, Ltd. (Osaka, Japan). Methods Preparation of CS-758 Solution and Suspension CS-758 solution was prepared by completely dissolving 10 mg of CS-758 drug substance with PEG 400/polysorbate 80 (1:1 [w/w]) solvent. CS-758 suspension was prepared by homogenously suspending 10 mg of CS-758 drug substance with 10 mL of 0.5% MC. Preparation of CS-758 Amorphous Solid Dispersion by Spray Drying The solubility of CS-758 is extremely low, and the rat bioavailability was approximately 53%.21 The amorphous solid dispersion has been developed as practical method to overcome the bioavailability of poorly water soluble drugs.22 The amorphous solid dispersion is the one of the active areas of research in the pharmaceutical field23 and is potential to enhance the bioavailability of CS-758. CS-758 amorphous solid dispersion was prepared by spray drying using CS-758 drug substance and a polymer to improve both the solubility and bioavailability. A physical mixture was prepared by mixing CS-758 with PVP K30 in a weight ratio of 1:1 using a mortar and a pestle. The physical mixture was dissolved in methanol. An amorphous solid dispersion composed of CS-758:PVP K-30 (1:1 (w/w)) was prepared using a Mini Spray Dryer B-290 (BÜCHI Labortechnik AG, Flawil, Switzerland). The inlet air temperature of 100 C, the exhaust air temperature of 60 C, the aspirator of 100%, and the spray rate of 15 mL/min were set as the spray drying conditions using B-290.
Powder X-ray Diffraction The powder X-ray diffraction (PXRD) patterns of the samples were measured using a Geiger Flex Rint-2200 diffractometer (Rigaku Corporation, Tokyo, Japan) with the CuKa radiation at 40 kV/40 mA. The samples were step-scanned at 0.02 intervals from 5.00 to 40.00 (2q) at the rate of 4.00 min1. The polymorph of CS-758 drug substance in the amorphous solid dispersion was characterized using PXRD. The PXRD patterns of unprocessed CS-758 drug substance, physical mixture blended with the drug substance and PVP K-30, and the amorphous solid dispersion (CS-758:PVP K-30 ¼ 1:1 [w/w]) are illustrated (Fig. S1). The unprocessed drug substance and the physical mixture showed sharp crystalline peaks. In contrast, the amorphous solid dispersion showed an amorphous halo pattern only. These results indicated that the polymeric structure of CS-758 drug substance in amorphous solid dispersion was amorphous. In Vitro Dissolution USP Apparatus 2 (Paddle Apparatus) described in USP <711> with UV-VIS analysis was used for in vitro dissolution testing. Each dissolution vessel was filled with 500 mL of dissolution medium of pH 6.9 phosphate buffer or pH 6.9 phosphate buffer with various concentration of SLS (0.1%, 0.15%, and 0.2% w/v), and the medium in the vessel was kept at 37 C. The stirring speed of paddle was set at 75 rpm. The dissolution testing (n ¼ 3) was conducted using 125 mg of CS-758 drug substance for all dosage forms. For generating the dissolution profiles, the dissolution test sample was taken from the vessel and filtered with a membrane filter (0.45 mm pore size) at 5, 10, 15, 20, 30, 45, 60, 90, 120, 150, 180, 210, 240, 270, 300, 330, and 360 min. The filtered sample was carried directly to the flow cell of the circulation system. The sample concentration was determined by UV-VIS spectrophotometry, and then the samples was returned to the vessel. In Vivo Animal Pharmacokinetic Experiments All animal experiments were performed in accordance with the in-house guidelines provided by the Institutional Animal Care and Use Committee of Daiichi Sankyo Company, Ltd. Male rats (n ¼ 3) were dosed in a nonrandomized, single-sequence design and fasted overnight. In a preparation of solution, 10 mg of CS-758 was completely dissolved with 4 mL of PEG 400/polysorbate 80 (1:1 [w/w]) solvent, and the saline was added up to 10 mL. In preparations of the suspension and the amorphous solid dispersion, 10 mg
1000
Plasma concentration (ng/mL)
3724
900
Observed
800
Simulated
700 600 500 400 300 200 100 0 0
3
6
9
12 15 Time (hr)
18
21
24
Figure 1. Observed and simulated rat plasma concentrationetime profiles after orally administered with CS-758 solution. The simulated profile (solid line) was obtained using ACAT model based on the physicochemical and pharmacokinetic properties, and the observed rat plasma concentrationetime profiles (closed diamond) after orally administered with CS-758 solution.
T. Kato et al. / Journal of Pharmaceutical Sciences 108 (2019) 3723-3728
3725
Table 1 Cmax, AUC0-t, and Tmax Predicted From a Simulated Rat Cp-Time Profile and Those Calculated From an Observed Rat Cp-Time Profile After Orally Administered With CS-758 Solution. %PEs of Cmax and AUC0-t Were Calculated Based on the Observed and Simulated Data Parameters
Observeda
Simulated
Absolute Percent Prediction Error (%PE)
Cmax (ng/mL) AUC0-t (ng/mL$hr) Tmax (hr)
703 7963 2
744 7907 2
6 1 -
%PE was evaluated as a reference of Guidance28 for Industry, Extended Release Oral Dosage Forms: Development, Evaluation, and Application of In Vitro/In Vivo Correlations. a Calculated by GastroPlus™.
of CS-758 or 20 mg of amorphous solid dispersion were suspended with 10 mL of 0.5% MC. Rat pharmacokinetic experiments were conducted under fasting conditions because bioavailability and bioequivalence studies for immediate release products are generally conducted under this condition. CS-758 solution or suspension were orally administered at a dose of 5 mg/5 mL/kg under fasting conditions. CS-758 amorphous solid dispersion was orally administered at a dose of 10 mg/5 mL/kg under fasting conditions. The dosing for these dosage forms was administered in a single oral dose via a catheter connected to a syringe into the stomach. After that, the syringe was exchanged to a new one filled with tap water and then the water was given to the rats. Blood samples were collected from jugular vein with a heparinized syringe at before administration (pre) or 0.5, 1, 2, 4, 8, and 24 h after administration. The collected blood was transferred to a microtube and put on ice immediately. The blood samples were centrifuged at 15,000 rpm for 5 min to separate the plasma. The supernatant was collected and stored at -20 C until analysis. The concentration of CS-758 in the supernatant was measured by liquid chromatography coupled with tandem mass spectrometry with the high-performance liquid chromatography system. In Silico Pharmacokinetic Modeling The rat Cp-time profiles for CS-758 were simulated using the commercially available software, GastroPlus™, version 9.0 (Simulations Plus Inc., Lancaster, CA). An ACAT model considering in vivo dissolution behaviors and gastrointestinal transit can be constructed using this software.24 An in silico PK model composed of an ACAT model was constructed using the physicochemical properties of CS-758 and the estimated PK parameter values. Input parameters used to construct the in silico PK model were set based on the in vitro experimental results and the values predicted by the ADMET predictor™. ADMET
predictor™ is commercially available software package to predict physicochemical and pharmacokinetic parameters based on structure formula of molecule.25 The input parameters used to construct the in silico PK model are summarized (Table S1). In the in silico PK model, the “IR: Solution” as dosage form option in Compound tab, and the “Rat-Physiological-Fasted” as Physiology and the “Opt logD Model SA/V 6.1” as ASF model in the Gut-physiology tab were selected. The input parameters of Log P, pKa, Peff (permeability parameter), blood/Cp ratio, and fraction unbound in plasma were predicted and set by ADMET predictor™. Clearance, volume of distribution, and the 2-compartment rate constants, k12 and k21, were calculated by fitting the simulated results to the observed rat Cp-time profile after orally administered with CS-758 solution using optimization module of GastroPlus™. Remaining model parameters used for this model were obtained as default setting parameters as shown (Table S1). For the integration of the in silico PK model into in vitro dissolution profiles, “CR: Dispersed” as dosage form option in Compound tab was selected. This option assumes that the unreleased drug is remained in gastrointestinal tract instead of disintegrating into individual particles and dispersing through the entire gastrointestinal tract. The simulation for rat Cp-time profiles using in vitro dissolution profiles were conducted using Weibull function in GastroPlus™. The release rate of the drug substance using Weibull function was calculated by the following equation.26,27
b A %Dose Released ¼ Max 1 f exp t Tlag where Max is percentage of maximum dissolution release, Tlag is lag time for dissolution, f is fraction of dissolution profile, t is time, A is the time scale factor, and b is the shape factor.
pH 6.9 phosphate buffer pH 6.9 phosphate buffer with 0.1%SLS pH 6.9 phosphate buffer with 0.15%SLS pH 6.9 phosphate buffer with 0.2%SLS
Plasma concentration (ng/mL)
Dissolution (%)
250 100 90 80 70 60 50 40 30 20 10 0
Observed Simulated(pH 6.9 phosphate buffer) Simulated(pH 6.9 phosphate buffer with 0.1%SLS) Simulated(pH 6.9 phosphate buffer with 0.15%SLS) Simulated(pH 6.9 phosphate buffer with 0.2%SLS)
200 150 100 50 0 0
0
30 60 90 120 150 180 210 240 270 300 330 360 Time (min)
Figure 2. Mean in vitro dissolution profiles (n ¼ 3) for CS-758 suspension using USP apparatus II in pH 6.9 phosphate buffer with various concentration of SLS (0%, 0.1%, 0.15%, and 0.2%). 500 mL of dissolution media was filled in vessels, and the media in the vessel were kept at 37 C. Stirring speed of paddle was set at 75 rpm. The dissolution testing was conducted using 125 mg of CS-758 drug substance for all dosage forms.
3
6
9
12 15 Time (hr)
18
21
24
Figure 3. Simulated rat plasma concentrationetime profiles after orally administered with CS-758 suspension using Weibull parameters fitted to the dissolution profiles in pH 6.9 phosphate buffer with various concentration of SLS (0%, 0.1%, 0.15%, and 0.2%). The rat plasma concentrationetime profiles of CS-758 suspension were predicted using the integrated in silico PK model into evaluated in vitro dissolution profiles as shown in Figure 2. The rat plasma concentration-time profiles of CS-758 suspension are shown as simulated profiles (solid lines) and an observed profile (closed diamond).
3726
T. Kato et al. / Journal of Pharmaceutical Sciences 108 (2019) 3723-3728
Evaluation of Prediction Accuracy for In Silico PK Model The absolute percent prediction error (%PE) was calculated by the following equation as the evaluation index for prediction accuracy of the established in silico PK model. The definition of %PE was referred from Guidance28 for Industry, Extended Release Oral Dosage Forms: Development, Evaluation, and Application of IVIVCs.
%PE ¼ ½ðObserved value Predicted valueÞ= Observed value 100 Based on the Guidance for Industry, the acceptance criteria of maximum plasma concentration (Cmax) and area under the plasma concentration-time curve (AUC0-t) predicted by the developed in silico PK model was established as 15% or less for %PE. Results and Discussion In Silico Pharmacokinetic Model Construction for CS-758 Solution The observed and simulated rat Cp-time profiles after orally administered with CS-758 solution were shown (Fig. 1). The Cmax, AUC0-t, and time to maximum plasma concentration (Tmax) calculated from observed and simulated Cp-time profiles were summarized (Table 1). As a result, the %PEs was calculated for Cmax (6%) and for AUC0-t (1%). The results indicated that the simulated Cmax and AUC0-t were reliable prediction accuracy because the %PEs for Cmax and AUC0-t met the predefined acceptance criterion, less than 15%. Therefore, the in silico PK model was judged to have a reliable prediction accuracy to predict the rat Cp-time profiles for CS-758 solution. Based on the result, the in silico PK model was considered to be applied for developing the in vitro dissolution method. Development of In Vitro Dissolution Method to Predict the Rat Plasma Concentration-Time Profiles of CS-758 Suspension To develop the in vitro dissolution method, the condition was explored to predict a rat Cp-time profile for CS-758 suspension. The mean dissolution profiles for CS-758 suspension using USP apparatus 2 in pH 6.9 phosphate buffer with various concentration of SLS (0%, 0.1%, 0.15%, and 0.2%) were shown (Fig. 2). The SLS is generally used as a surfactant to accomplish the desired dissolution profiles for poorly water-soluble compounds. The SLS was added to the dissolution medium for the in vitro dissolution method development because the solubility of CS-758 was extremely low (0.05 mg/mL in phosphate buffer) as shown (Table S1). The results indicated that an in vitro dissolution profile of CS-758 suspension in pH 6.9 phosphate buffer without SLS was extremely low (approximately 1% of dissolution rate at 360 min). The dissolution rate was increased with the addition of SLS, the more SLS concentrations the higher dissolution
rate. The dissolution rate at 360 min was approximately 23% when the SLS concentration in the dissolution medium was 0.2%. Based on the in vitro dissolution results, the obtained in vitro dissolution profiles were incorporated into the constructed in silico PK model to predict a rat Cp-time profile for CS-758 suspension. Weibull parameters were determined by fitting to the mean in vitro dissolution profiles using GastroPlus™ software. The determined Weibull parameters with various concentration of SLS in pH 6.9 phosphate buffer are summarized (Table S2). These determined Weibull parameters were used as input parameters for the in silico PK simulation. Then, the in vitro dissolution condition with the best prediction accuracy was selected based on the result of %PE evaluation. The observed and simulated rat mean Cp-time profiles after orally administered with CS-758 suspension are shown (Fig. 3). The %PEs for the Cmax and AUC0-t for CS-758 suspension evaluated using GastroPlus™ were summarized (Table 2). The %PEs were calculated for Cmax (4%) and AUC0-t (0.5%) when the in vitro dissolution profile in pH 6.9 phosphate buffer with 0.15% SLS concentration was incorporated into the constructed in silico PK model. On the other hand, the %PEs for the Cmax and AUC0-t were not reliable prediction accuracy (more than 30% of %PEs) when the in vitro dissolution profiles in pH 6.9 phosphate buffer with the other SLS concentration were incorporated into the constructed in silico PK model. The results indicated that %PEs for the Cmax and AUC0-t calculated using the dissolution profile of pH 6.9 phosphate buffer with 0.15% SLS met the predefined acceptance criterion, less than 15%. Based on the results, USP Apparatus 2 (Paddle Apparatus) of paddle rotation speed of 75 rpm and 500 mL of pH 6.9 phosphate buffer with 0.15% SLS was applied as a dissolution method that can predict the rat Cp-time profile of CS-758 suspension. The dissolution of CS-758 drug substance in the suspension with this in vitro dissolution condition did not show the complete dissolution because the dissolution rate at 360 min was low (approximately 16%). However, the integration of in silico PK model into this dissolution profile had the desirable prediction accuracy to reflect the rat Cp-time profile of CS-758 suspension. Therefore, the in vivo dissolution behavior of the CS-758 suspension in the intestine was also considered to indicate an incomplete dissolution behavior because CS-758 drug substance was of low solubility. Moreover, in this investigation, an addition of SLS in the dissolution medium was considered to be optimal to predict the rat Cp-time profile of CS-758 suspension although the SLS is not included as the components in gastric and intestinal fluids. The predictable in silico PK model could be developed by an integration of the in silico model into in vitro dissolution profiles of CS-758 suspension because this model was confirmed to have the desirable prediction accuracy. These results suggest that the integration of in silico PK model into in vitro dissolution profiles can be applied to predict the rat Cp-time profiles for the other formulation (i.e. amorphous solid dispersion), with different in vitro dissolution behaviors.
Table 2 Cmax, AUC0-t, and Tmax Predicted From Simulated Rat Cp-Time Profiles Using In Vitro Dissolution Data in pH 6.9 Phosphate Buffer With Various Concentration of SLS and Those Calculated From an Observed Cp-Time Profile After Orally Administered With CS-758 Suspension. %PEs of Cmax and AUC0-t Were Calculated Based on the Observed and Simulated Data Parameters
Cmax (ng/mL) AUC0-t (ng/ mL$hr) Tmax (hr)
Observeda Without SLS
0.1% SLS
0.15% SLS
0.2% SLS
Simulated Absolute Percent Prediction Error (%PE)
Simulated Absolute Percent Prediction Error (%PE)
Simulated Absolute Percent Prediction Error (%PE)
Simulated Absolute Percent Prediction Error (%PE)
103
7.9
92
46
55
99
4
136
32
1315
120
91
638
51
1309
0.5
1903
45
4
4
-
4
-
4
-
4
-
%PE was evaluated as a reference of Guidance a Calculated by GastroPlus™.
28
for Industry, Extended Release Oral Dosage Forms: Development, Evaluation, and Application of In Vitro/In Vivo correlations.
Dissolution (%)
T. Kato et al. / Journal of Pharmaceutical Sciences 108 (2019) 3723-3728
100 90 80 70 60 50 40 30 20 10 0
CS-758 suspension CS-758 amorphous solid dispersion (CS-758:PVP K-30 =1:1 (w/w))
0
30 60 90 120 150 180 210 240 270 300 330 360 Time (min)
Figure 4. Dissolution profiles (n ¼ 3) of CS-758 suspension (cross) and amorphous solid dispersion (closed circle) in pH 6.9 phosphate buffer with 0.15% SLS. 500 mL of dissolution medium was filled in vessels and the medium in the vessel was kept at 37 C. Stirring speed of paddle was set at 75 rpm. The dissolution testing was conducted using 125 mg of CS-758 drug substance for all dosage forms.
Verification of a Developed In Silico PK Model by Predicting Rat Plasma Concentration-Time Profiles of CS-758 Amorphous Solid Dispersion In this investigation, the developed in silico PK modeling approaches that were integrated into in vitro dissolution method were further evaluated as a verification process by predicting a rat Cptime profile for CS-758 amorphous solid dispersion. The in vitro dissolution profiles of the CS-758 suspension and amorphous solid dispersion in pH 6.9 phosphate buffer with 0.15% SLS concentration were shown (Fig. 4). The result showed that the in vitro dissolution profile of CS-758 amorphous solid dispersion was higher than that of CS-758 suspension. The in vitro dissolution rate of CS-758 was considered to be increased by an effect of the amorphous state of CS-758 drug substance in the amorphous solid dispersion. Based on the in vitro dissolution results, the obtained in vitro dissolution profiles were incorporated into the constructed in silico PK model to predict a rat Cp-time profile for CS-758 amorphous solid dispersion. Weibull parameters were determined by fitting to a mean in vitro dissolution profile using GastroPlus™ software. The determined Weibull parameters are summarized (Table S3). The observed and simulated rat mean Cp-time profiles of CS-758 suspension and amorphous solid dispersion are illustrated (Fig. 5). The simulated Cmax and AUC0-t for CS-758 amorphous solid dispersion were higher than those for the CS-758 suspension when the in vitro dissolution profile for CS-758 amorphous solid dispersion was
incorporated into the constructed in silico PK model (approx. 3 times higher) (Table 2 vs. Table 3). The results indicated that the bioavailability of the amorphous solid dispersion would be predicted to be improved in comparison with that of the CS-758 suspension. The %PEs for the Cmax and AUC0-t for CS-758 amorphous solid dispersion evaluated using GastroPlus™ were summarized (Table 3). A result indicated that the simulated Cp-time profile for the amorphous solid dispersion was comparable to the observed one because the %PEs for the Cmax (9%) and AUC0-t (11%) met the predefined acceptance criterion, less than 15%. Therefore, it was demonstrated that the in silico PK model integrated into in vitro dissolution profiles had a reliable prediction accuracy for rat Cptime profiles for suspension and amorphous solid dispersion. The predictable in silico PK model could be verified by an integration of the model into an in vitro dissolution profile of CS-758 amorphous solid dispersion because this in silico model was also confirmed to have the desirable prediction accuracy. This in silico modeling approaches were judged to be predictable approaches to predict in vivo PK behaviors by development and verification processes. It has been studied that an IVIVC approach using an in vitro dissolution profiles and in silico PK model could predict PK behaviors of modified release formulation for BCS class I compound.19 Moreover, it was reported that a mechanistic-based IVIVC approach using an in vitro dissolution profiles and in silico PK model for BCS class III compound was developed for demonstrating acceptable Cp-time profiles. A strategy for considering biowaiver approach was provided by this mechanistic-based approach.20 These reports concluded that IVIVC approaches using an integration of in silico PK model into in vitro dissolution profiles were applicable to simulate Cp-time profiles for water soluble compounds (e.g. BCS class I and III). Based on these reports, the investigational results in this section supported that the integration of in silico PK model into in vitro dissolution profiles can also be useful for formulation screening considering PK behaviors for poorly soluble compounds (e.g. BCS class II or IV). Therefore, the integration of in silico PK modeling approaches will be considered to be applicable for the other poorly soluble compounds, depending on the physicochemical properties and PK behaviors. The conventional IVIVC relationship should be demonstrated consistently with 2 or more formulations with different release rates to result in corresponding differences in absorption profiles. Although an IVIVC can be defined with a minimum of 2 formulations with different release rates, 3 or more formulations with different release rate are recommended.28 On the other hand, the integration of in silico PK modeling approaches into in vitro dissolution profiles could be constructed based on one dosage form with one in vitro dissolution profile. Therefore, the in silico PK modeling
500
Plasma concentration (ng/mL)
3727
Observed (CS-758 suspension)
450 400 350
Simulated (CS-758 suspension)
300 250
Observed (CS-758 amorphos solid dispersion (CS-758:PVP K-30=1:1 (w/w))
200 150
Simulated (CS-758 amorphos solid dispersion (CS-758:PVP K-30=1:1 (w/w))
100 50 0 0
3
6
9
12 15 Time (hr)
18
21
24
Figure 5. Simulated (plots) and observed (solid lines) rat plasma concentrationetime profiles after orally administered with CS-758 suspension and amorphous solid dispersion. The rat plasma concentrationetime profiles of CS-758 suspension and amorphous solid dispersion were predicted using the integrated in silico PK model into evaluated in vitro dissolution profiles as shown in Figure 4. The rat plasma concentrationetime profiles of CS-758 suspension and amorphous solid dispersion are shown as simulated profiles (blue and orange solid lines) and an observed profile (closed diamond and cross).
3728
T. Kato et al. / Journal of Pharmaceutical Sciences 108 (2019) 3723-3728
Table 3 Cmax, AUC0-t, and Tmax Predicted From a Simulated Rat Cp-Time Profile and Those Calculated From an Observed Rat Cp-Time Profile After Orally Administered With CS-758 Amorphous Solid Dispersion. %PEs of Cmax and AUC0-t Were Calculated Based on the Observed and Simulated Data Parameters
Observeda
Simulated
Absolute Percent Prediction Error (%PE)
Cmax (ng/mL) AUC0-t (ng/mL$hr) Tmax (hr)
299 3893 4
267b 4239c 5
11 9 -
%PE was evaluated as a reference of Guidance28 for Industry, Extended Release Oral Dosage Forms: Development, Evaluation, and Application of In Vitro/In Vivo Correlations. a Calculated by GastroPlus™. b 2.7 times for simulated Cmax for CS-758 suspension. c 3.2 times for simulated AUC0-t for CS-758 suspension.
approaches can construct with less in vitro dissolution profile data compared with the conventional IVIVC approaches. The in silico PK modeling approaches can be useful for finding optimal formulation to accomplish target product profiles (target dissolution profiles) and target PK behaviors with the desirable prediction accuracy. Conclusions The investigation of stepwise integration of in silico PK modeling approaches into in vitro dissolution profiles was demonstrated to reproduce the observed PK profiles for oral immediate release drug products (i.e. solution, suspension, and amorphous solid dispersion). The in silico PK model was constructed using observed PK data for CS-758 solution. In addition, an in vitro dissolution method was developed to predict a rat Cp-time profile for suspension. Furthermore, the PK profile of CS-758 amorphous solid dispersion was accurately predicted based on the %PE metrics. These approaches can contribute to the formulation screening for poorly soluble compounds (e.g. BCS class II or IV) by predicting PK behaviors with the dissolution profile, even if the in vitro dissolution methods are pharmacopeially applied as a quality control tools. Same approaches can be applied and useful for human PK behaviors using observed human Cp-time profiles. Acknowledgments € hling The authors would like to acknowledge Pearnchob-Ho Nantharat, Dr. and Hiroshi Nakagawa, Dr. in Daiichi Sankyo Europe GmbH, and Tsuyoshi Mikkaichi in Daiichi Sankyo Co., Ltd. for valuable consultations regarding preparation for this scientific article. The authors would like to acknowledge Yumiko Fujii for valuable consultations regarding submitting this scientific article. References 1. Dressman JB, Reppas C. In vitroein vivo correlations for lipophilic, poorly water-soluble drugs. Eur J Pharm Sci. 2000;11:S73-S80. 2. Balan G, Timmins P, Greene DS, Marathe PH. In-vitro in-vivo correlation models for glibenclamide after administration of metformin/glibenclamide tablets to healthy human volunteers. J Pharm Pharmacol. 2000;52:831-838. 3. Nicolaides E, Symillides M, Dressman JB, Reppas C. Biorelevant dissolution testing to predict the plasma profile of lipophilic drugs after oral administration. Pharm Res. 2001;18:380-388.
4. Balan G, Timmins P, Greene DS, Marathe PH. In vitroein vivo correlation (IVIVC) models for metformin after administration of modified-release (MR) oral dosage forms to healthy human volunteers. J Pharm Sci. 2001;90:1176-1185. 5. Sirisuth N, Augsburger LL, Eddington ND. Development and validation of a nonlinear IVIVC model for a diltiazem extended release formulation. Biopharm Drug Dispos. 2002;23:1-8. 6. Sunesen VH, Pedersen BL, Kristensen HG, Mullertz A. In vivo in vitro correlations for a poorly soluble drug, danazol, using the flow-through dissolution method with biorelevant dissolution media. Eur J Pharm Sci. 2005;24:305-313. 7. Ghosh A, Bhaumik UK, Bose A, et al. Extended release dosage form of glipizide: development and validation of a level A in vitroein vivo correlation. Biol Pharm Bull. 2008;31:1946-1951. 8. Macha S, Yong CL, Darrington T, et al. In vitroein vivo correlation for nevirapine extended release tablets. Biopharm Drug Dispos. 2009;30:542-550. 9. Rossi RC, Dias CL, Bajerski L, et al. Development and validation of discriminating method of dissolution for fosamprenavir tablets based on in vivo data. J Pharm Biomed Anal. 2011;54:439-444. 10. Saibi Y, Sato H, Tachiki H. Developing in vitroein vivo correlation of risperidone immediate release tablet. AAPS PharmSciTech. 2012;13:890-895. 11. Guhmann M, Thommes M, Gerber F, et al. Design of biorelevant test setups for the prediction of diclofenac in vivo features after oral administration. Pharm Res. 2013;30:1483-1501. 12. Kesisoglou F, Rossenu S, Farrell C, et al. Development of in vitro-in vivo correlation for extended-release niacin after administration of hypromellosebased matrix formulations to healthy volunteers. J Pharm Sci. 2014;103: 3713-3723. 13. Li S, He H, Parthiban LJ, et al. IV-IVC considerations in the development of immediate-release oral dosage form. J Pharm Sci. 2005;94:1396-1417. 14. Bredael GM, Bowers N, Boulineau F, et al. In vitroein vivo correlation strategy applied to an immediate-release solid oral dosage form with a Biopharmaceutical Classification System IV compound case study. J Pharm Sci. 2014;103:2125-2130. 15. Lagece M, Gravelle L, Di Maso M, et al. Developing a discriminating dissolution procedure for a dual active pharmaceutical product with unique solubility characteristics. Dissolution Technol. 2004;11:13-17. 16. He Z, Zhong D, Chen X, et al. Development of a dissolution medium for nimodipine tablets based on bioavailability evaluation. Eur J Pharm Sci. 2004;21:487-491. 17. Wu C, Liu Y, He Z, et al. Insight into the development of dissolution media for BCS Class II drugs: a review from quality control and prediction of in vivo performance perspectives. Curr Drug Deliv. 2016;13:1004-1020. 18. Kostewicz ES, Aarons L, Bergstrand M, et al. PK models for the prediction of in vivo performance of oral dosage forms. Eur J Pharm Sci. 2014;57:300-321. 19. Mirza T, Bykadi SA, Ellison CD, et al. Use of in vitroein vivo correlation to predict the pharmacokinetics of several products containing a BCS class 1 drug in extended release matrices. Pharm Res. 2013;30:179-190. 20. Crison JR, Timmins P, Keung A, et al. Biowaiver approach for biopharmaceutics Classification system class 3 compound metformin hydrochloride using in silico modeling. J Pharm Sci. 2012;101:1773-1782. 21. Kagoshima Y, Mori M, Suzuki E, et al. Synthesis, cleavage, and antifungal activity of a number of novel, water-soluble ester prodrugs of antifungal triazole CS-758. Biroog Med Chem. 2009;19:3559-3563. 22. Sekiguchi K, Obi N. Studies on absorption of eutectic mixture. I. A comparison of the behavior of eutectic mixture of sulfathiazole and that of ordinary sulfathiazole in man. Chem Pharm Bull. 1961;9:866-872. 23. Serajuddin ATM. Solid dispersion of poorly water-soluble drugs: early promises, subsequent problems, and recent breakthroughs. J Pharm Sci. 1999;88: 1058-1066. 24. Agoram B, Woltosz WS, Bolger MB. Predicting the impact of pharmacokinetic and biochemical processes on oral drug bioavailability. Adv Drug Deliv Rev. 2001;50:S41-S67. 25. Emoto C, Murayama N, Rostami-Hodjegan A, et al. Utilization of estimated physicochemical properties as an integrated part of predicting hepatic clearance in the early drug-discovery stage: impact of plasma and microsomal binding. Xenobiotica. 2009;39:227-235. 26. Wang B, Liu Z, Li D, et al. Application of physiologically based pharmacokinetic modeling in the prediction of pharmacokinetics of bicyclol controlled-release formulation in human. Eur J Pharm Sci. 2015;77:265-272. 27. Babiskin AH, Zhang X. Application of pharmacokineticly based absorption modeling for amphetamine salts drug products in generic drug evaluation. J Pharm Sci. 2015;104:3170-3182. 28. FDA. Guidance for industry. Extended Release Oral Dosage Forms: Development, Evaluation and Application of in Vitro/in Vivo Correlations. Rockville (MD): Center for Drug Evaluation and Research (CDER), US Department of Health and Human Services; 1997.