Prediction of Ketoconazole absorption using an updated in vitro transfer model coupled to physiologically based pharmacokinetic modelling

Accepted Manuscript Prediction of Ketoconazole absorption using an updated in vitro transfer model coupled to physiologically based pharmacokinetic modelling

Aaron Ruff, Tom Fiolka, Edmund S. Kostewicz PII: DOI: Reference:

S0928-0987(16)30557-7 doi: 10.1016/j.ejps.2016.12.017 PHASCI 3834

To appear in:

European Journal of Pharmaceutical Sciences

Received date: Revised date: Accepted date:

19 September 2016 14 December 2016 19 December 2016

Please cite this article as: Aaron Ruff, Tom Fiolka, Edmund S. Kostewicz , Prediction of Ketoconazole absorption using an updated in vitro transfer model coupled to physiologically based pharmacokinetic modelling. The address for the corresponding author was captured as affiliation for all authors. Please check if appropriate. Phasci(2016), doi: 10.1016/j.ejps.2016.12.017

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT Prediction of Ketoconazole absorption using an updated in vitro transfer model coupled to physiologically based pharmacokinetic modelling

Aaron Ruff, Tom Fiolka and Edmund S. Kostewicz

PT

Institute of Pharmaceutical Technology, Goethe University, Frankfurt/Main, Germany

RI

Corresponding Author

SC

Edmund S Kostewicz Institute of Pharmaceutical Technology,

NU

Goethe University,

MA

Max-von-Laue Str. 9

AC

CE

PT E

D

60438 Frankfurt/Main, Germany

1

ACCEPTED MANUSCRIPT Abbreviations: Area under the curve (AUC) Biopharmaceutical classification system (BCS) Critical supersaturation concentration (CSC)

PT

Degree of supersaturation (DOS) Fasted state simulated gastric fluid (FaSSGF)

RI

Fasted state simulated gastric fluid version 2 high bile salt concentration (FaSSGF-V2-HBS)

SC

Fasted state simulated intestinal fluid (FaSSIF)

NU

Fasted state simulated intestinal fluid version 2 high buffer capacity (FaSSIF-V2-HBC) Fraction dissolved (fd)

MA

Fraction solid (fs)

In vitro in vivo correlation (IVIVC)

PT E

Ketoconazole (KTZ)

D

Gastrointestinal tract (GIT)

Maximal concentration (Cmax)

CE

Migrating Motor Complex (MMC)

AC

Pharmacokinetic (PK)

Physiologically based pharmacokinetic modelling (PBPK) Precipitation rate constant (PRC) Rounds per minute (rpm) Simulated intestinal fluids (SIF) Time at which maximal concentration is observed (Tmax) Volume of distribution (Vd)

2

ACCEPTED MANUSCRIPT Key words

Physiologically-based pharmacokinetic (PBPK) model; Predicting drug absorption; in vitro in

AC

CE

PT E

D

MA

NU

SC

RI

PT

vivo correlation (IVIVC); Supersaturation; Precipitation; Transfer model; Ketoconazole

3

ACCEPTED MANUSCRIPT Abstract The aim of this study was to optimize the in vitro transfer model and to increase its biorelevance to more accurately mimic the in vivo supersaturation and precipitation behaviour of weak basic drugs. Therefore, disintegration of the formulation, volumes of the stomach and intestinal compartments, transfer rate, bile salt concentration, pH range and paddle speed were varied over a physiological relevant range. The supersaturation and precipitation data from these experiments for ketoconazole

PT

(KTZ) were coupled to physiologically based pharmacokinetic (PBPK) model using Stella® software, which also incorporated the disposition kinetics of KTZ taken from the literature, in order to simulate

RI

the oral absorption and plasma profile in humans. As expected for a poorly soluble weak base, KTZ demonstrated supersaturation followed by precipitation under various in vitro conditions simulating

SC

the proximal small intestine with the results influenced by transfer rate, hydrodynamics, volume, bile salt concentration and pH values. When the in vitro data representing the “average” GI conditions

NU

was coupled to the PBPK model, the simulated profiles came closest to the observed mean plasma profiles for KTZ. In line with the high permeability of KTZ, the simulated profiles were highly influenced by supersaturation whilst precipitation was not predicted to occur in vivo. A physiological

MA

relevant in vitro “standard” transfer model setup to investigate supersaturation and precipitation was established. For translating the in vitro data to the in vivo setting, it is important that

AC

CE

PT E

D

permeability is considered which can be achieved by coupling the in vitro data to PBPK modelling.

4

ACCEPTED MANUSCRIPT 1. Introduction The number of water insoluble new drug candidates progressing through pharmaceutical development over the past decade has increased considerably (Lobell et al., 2006). In order to overcome solubility limitations, a formulation approach to enhance the apparent concentration of drug in the gastrointestinal (GI) lumen can be achieved through supersaturation (Kostewicz et al., 2014). A greater amount of drug in solution means that more drug is available for absorption across

PT

the intestinal mucosa. Alternatively, supersaturation may also result from the transfer of a weak base (irrespective of the formulation used) from an acidic stomach into the more pH neutral small

RI

intestine. In both cases, the supersaturated solution is thermodynamically unstable and the drug may precipitate. Supersaturation and precipitation along the GI tract may therefore have a significant

SC

impact on the overall absorption of poorly soluble drugs administered via the oral route. Whilst solubility is a key parameter that needs to be investigated during product development, the

NU

solubility data do not describe the supersaturation and precipitation characteristic of the drug. The dissolution behaviour of the formulation is also important to consider, but given that the

MA

conventional USP1 and USP2 apparatus typically utilize a single media and volume at constant pH, dissolution results also do not adequately reflect the in vivo situation, in which transit through the GI

D

tract exposes the drug/formulation to a constantly changing environment (Kostewicz et al., 2002). As supersaturation and precipitation can be influenced by a multitude of factors, these need to be

PT E

considered in any in vitro model used. For example, supersaturation concentrations in the lumen may be influenced by gastric emptying, ionization concentrations of the drug, solubilisation by bile acid micelles and dissolution characteristics of the formulation. Factors influencing precipitation

CE

along the GI tract include the pH transition between the stomach and proximal intestine, dilution of the formulation by GI fluids and corresponding composition of the GI fluids, and characteristics of the

AC

excipients used in the formulation (Brouwers et al., 2009; Tonsberg et al., 2010; Xu and Dai, 2013). To capture these parameters appropriately, the transfer model, which was initially presented in 2004 (Kostewicz et al., 2004), is a multi-compartmental model taking into account both the stomach and intestinal compartments and has been frequently used to investigate the supersaturation and precipitation characteristics for poorly soluble weak bases (Berlin et al., 2014; Kostewicz et al., 2004; Wagner et al., 2012; Xu and Dai, 2013). To ensure that the experimental conditions of the transfer model appropriately captures the relevant GI physiological, the following parameters were investigated: disintegration of the formulation, volume of the stomach and intestinal compartments, transfer rate, bile salt concentration, pH range and paddle speed. To evaluate the impact of varying the physiological parameters, the results from 5

ACCEPTED MANUSCRIPT the in vitro transfer experiments were coupled to an updated Stella® PBPK model and the resulting simulated plasma profile compared to literature plasma profiles. The model drug chosen to optimise the transfer model was Ketoconazole (KTZ), a BCS II antifungal agent, which exhibits a diphasic pKa (6.5 and 2.9) (Blum et al., 1991) and a log P value of 3.9 (Ghasemi and Saaidpour, 2007). Further, given the availability of in vivo data with respect to luminal precipitation (Psachoulias et al., 2011) and human pharmacokinetic data after administration of

PT

different formulations of KTZ (Huang et al., 1986), enabled an in vitro in vivo correlation (IVIVC) to be

AC

CE

PT E

D

MA

NU

SC

RI

established.

6

ACCEPTED MANUSCRIPT 2. Materials and Methods 2.1. Chemicals and reagents KTZ powder was purchased from Caesar & Lorentz GmbH (Hilden, Germany). Nizoral® tablets containing 200 mg of KTZ were kindly donated by Janssen Pharmaceutics (Buckinghamshire, UK). Sodium hydroxide was purchased from Merck KGaA (Darmstadt, Germany). Sodium chloride, orthophosphoric acid and hydrochloric acid 37% were purchased from VWR International (Leuven,

PT

Belgium) and maleic acid was purchased from AppliChem GmbH (Darmstadt, Germany). Organic solvents for HPLC analysis including trimethylamine and acetonitrile were all HPLC grade and

RI

purchased from Merck-Schuchardt (Hohebrunn, Germany) and Merck KGaA (Darmstadt, Germany),

SC

respectively. SIF powder original and SIF powder V2 were a kind donation from biorelevant.com

NU

(London, UK). All other chemicals were analytical grade or equivalent, and purchased commercially.

2.2. Media used for solubility, dissolution and transfer experiments

MA

2.2.1. Composition of media

Compendial and biorelevant media were used to evaluate the solubility, dissolution, supersaturation

D

and precipitation characteristics of KTZ. The biorelevant media were prepared using SIF-powders based on instructions from www.biorelevant.com. Detailed information on the composition of all

PT E

compendial and biorelevant media used in this study is presented in Table S1 (see Supplementary Information).

CE

To simulate the conditions in the stomach, FaSSGF-V2 (Vertzoni et al., 2007) was used at pH 2.0 rather than at pH 1.6 (Fei et al., 2013), in order to reduce the drop in pH during transfer into the intestinal compartment during the transfer experiments. Additionally, to reduce the drop in pH of

AC

the intestinal compartment following transfer of the gastric media and to maintain the pH in the acceptor compartment at approximately 5.8, FaSSIF-V2 with a higher buffer capacity was used (Maleate(HBC) / FaSSIF-V2(HBC)). To maintain the bile salt concentration of FaSSIF-V2 following transfer of FaSSGF-V2 into the intestinal compartment, for some experiments the bile salt concentration in FaSSGF-V2 was increased to the same concentration as used in FaSSIF-V2 (FaSSGFV2(HBS)). FaSSIF-V1 was prepared using the same buffer composition which was used for FaSSIF-V2 (i.e. maleate buffer). To replicate the dilution of bile salts and reduction in pH in the intestinal compartment after transfer of the gastric compartment in the transfer experiment and it evaluate its impact on solubility, various ratios of the gastric to intestinal media including FaSSGF-V2 (pH 2.0) with FaSSIF-V2 (ratio 1:2 and 1:1.4), FaSSGF-V2 (pH 2.0) with FaSSIF-V1 (ratio 1:1.4), FaSSGF-V2 (pH 7

ACCEPTED MANUSCRIPT 2.0) with FaSSIF-V2(HBc) (ratio 2.5:1) and FaSSGF-V2(HBS) with FaSSIF-V2 (ratio 1:1.4) were prepared.

2.3. Solubility experiments To evaluate the solubility of KTZ in the gastric and intestinal compartments, experiments were

PT

executed using both compendial and biorelevant media. To represent the stomach HCl / NaCl buffers and FaSSGF-V2 at pH values of 1.0, 1.6 and 2.0, and FaSSGF-V2(HBS) were used. To simulate the

RI

intestine before and after transfer of the gastric compartment, maleate buffer, FaSSIF-V1, FaSSIF-V2, FaSSIF-V2(HBc) and the appropriate ratios of gastric to intestinal media as described in section 2.2.1.

SC

were prepared.

The solubility of KTZ was evaluated using the the Uniprep™ system as previously described by

NU

Glomme et al. (Glomme et al., 2005). After incubation at 37°C for 24 h, the concentration of KTZ was

MA

measured by HPLC (section 2.5). All solubility experiments were performed in triplicate.

2.4. Dissolution and transfer experiments

D

2.4.1. Equipment

PT E

The dissolution and transfer experiments were performed using an ERWEKA® DT 600 USP II dissolution tester, utilizing three 500 mL mini-vessels (stomach compartment) and three standard 1000 mL vessels (intestinal compartment), each with the respective corresponding mini-paddle or

AC

CE

standard paddle system setup (Erweka® GmBH, Heusenstamm, Germany).

2.4.2. Experimental setup Dissolution:

The dissolution characteristics of KTZ (Nizoral® tablet) were evaluated in both 250 mL of FaSSGF-V2 and 500 mL of FaSSIF-V2. Samples were taken out to 240 minutes. Experiments were performed in triplicate using an agitation speed of 100 rpm at 37 °C ± 0.5 °C. Transfer model: The in vitro transfer model comprises of a two compartment setup simulating the stomach (donor) and intestinal (acceptor) compartment utilizing a mini-vessel / paddle and standard vessel / paddle 8

ACCEPTED MANUSCRIPT setup, respectively (Kostewicz et al., 2004). To simulate the transfer of gastric contents (including dissolved and undissolved particles) into the small intestine, a programmable ISMATEC® MC-Process IP5 peristaltic pump was used. For each donor and acceptor compartment pair, a 70-80 cm length of tubing (ISMATEC® neoprene) with an internal diameter of 2.06 mm was used to allow the transfer of the gastric to the intestinal compartment (IDEX Health & Science GmbH, Wertheim Germany). To position the tubing at the bottom of the mini-vessel, a glass tubing system with a length of 11 cm and a diameter of approximately 0.6 mm (I.D. 0.5 mm) was securely fixed on the dissolution vessel lid,

PT

through which the neoprene tubes were threaded into the donor compartment (fig. 1). For the intestinal compartment, a similar glass cylinder with a length of 6.5 cm was positioned in the

RI

dissolution vessel lid through which the neopren tubing was thread and positioned just below the

AC

CE

PT E

D

MA

NU

SC

surface of the intestinal media.

Fig. 1. Graphical representation of the in vitro transfer model setup

2.4.3. Detailed evaluation of the transfer model setup As part of the update of the transfer model, not only the disintegration of the formulation in the gastric compartment, but also the range of GI physiology with respect to relevant GI volumes, transfer rates, bile salt concentrations and hydrodynamics observed along the proximal GI tract were taken into consideration. To begin with, the “standard” transfer model conditions were identified to represent the “average” physiology and using these experimental conditions, the influence of each 9

ACCEPTED MANUSCRIPT parameter was evaluated in isolation (i.e. altering the parameter in question but keeping the other parameters fixed) (table 1). Disintegration of the formulation: To evaluate the influence of disintegration and dissolution of the formulation in the stomach compartment and the subsequent transfer of undissolved drug particles into the intestinal compartment, experiments were performed using either predissolved Nizoral® tablets in the donor

PT

compartment or by placing the formulation into the donor compartment and immediately beginning

RI

the transfer.

SC

Volume of the “fasting” GI compartment:

To more closely reflect the fasting volumes in the upper GI tract, the gastric compartment was nominally set to 250 mL (Goetze et al., 2009; Gruber et al., 1987; Mudie et al., 2014). To get a better

NU

understanding of the impact of intestinal volume on the supersaturation and precipitation characteristics of KTZ, and to consider the range of intestinal volumes that can be anticipated (e.g.

MA

40-320 mL), experiments were conducted using 100 and 500 ml as the extremes and a 350 ml volume was nominally set as the standard volume representing the proximal intestine (Marciani et

D

al., 2010; Schiller et al., 2005; Sutton, 2009).

PT E

Transfer rate:

Physiologically relevant first order transfer rates (Hunt, 1956; Moore et al., 1984; Steingoetter et al., 2006) with half-life times of 22.4 min (Oberle et al., 1990) and 9 min (Adkin et al., 1995; Murray et al.,

CE

1994; Oberle et al., 1990; Steingoetter et al., 2006; Wilding et al., 1994) were compared to 2 mL/min and 9 mL/min zero order transfer. The intermediate 9 min half-life was identified as the transfer rate to reflect the average physiology of the transfer model. To evaluate the influence of a very rapid

AC

gastric emptying, dumping experiments (i.e. predissolved KTZ in the donor compartment was manually poured into the intestinal compartment over a period of 15 seconds) were conducted and compared to the first and zero order transfer rates. Bile salt concentration: Given that a high range of bile salt concentration has been reported in the literature for the small intestine (Annaert et al., 2010; Bevernage et al., 2010; Brouwers et al., 2006; Clarysse et al., 2009a; Clarysse et al., 2009b), a series of experiments were conducted to evaluate the impact of bile salt concentration on the supersaturation and precipitation characteristics of KTZ. The transfer experiments were conducted using, (1) compendial media (HCl/NaCL) buffer at pH 2.0; Maleate 10

ACCEPTED MANUSCRIPT buffer in the absence of bile salts, (2) the standard experimental setup where the bile salt concentration in FaSSIF-V2 is diluted from 3.0 mM to 1.78 mM, (3) FaSSIF-V1, which has a slightly higher initial bile salt concentration than FaSSIF-V2 ,and (4) using FaSSGF-V2(HBC) where the gastric media contained bile salts at the same concentration as in FaSSIF-V2 in order to maintain the bile salt in the acceptor compartment. Hydrodynamics / Paddle speed:

PT

To evaluate the impact of hydrodynamics on the disintegration and dissolution behaviour of the formulation in the donor compartment, and on the supersaturation and precipitation behaviour of

RI

KTZ, transfer experiments using agitation speeds of 50, 100 and 150 rpm were performed. For the

SC

standard transfer model, the paddle speed was set to 100 rpm. Experiments were performed using Nizoral® tablets which were either predissolved or undergoing simultaneous dissolution in the donor

NU

compartment during transfer into the acceptor compartment.

Table 1: Summary of the experimental conditions used to evaluate the transfer model Standard setup

MA

Parameter

(based on average physiology)

Range investigated (physiological “extreme” conditions)

Simultaneous dissolution in gastric Predissolved formulation in gastric

formulation

compartment and transfer

Gastric conditions

250 mL FaSSGF-V2 (pH 2.0)

Intestinal volume

350 mL

100 mL – 500 mL

FaSSIF-V2

Blank buffer – FaSSIF-V1

Taurocholate: 3.0 mM  1.75mM*

Taurocholate: 0.0 mM  3.0 mM

PT E

D

Disintegration of

CE

Intestinal media

Lecithin:

AC

Gastric emptying

Paddle speed

Sampling points in the

0.2 mM  0.12 mM*

First order: t ½ = 9 min

compartment

Lecithin:

Zero order: 2 mL/min – 9 mL/min First order: t ½ = 22.4 min Special:

100 rpm

0.0 mM  0.44 mM

gastric dumping

50 rpm – 150 rpm

2, 5, 10, 15, 20, 30, 45, 60, 75, 90, 120, 180, 240 min

intestinal compartment Temperature

37 ± 0.5 °C

* Reduction of bile salt concentration during transfer experiment

11

ACCEPTED MANUSCRIPT 2.4.4. Sampling and sample preparation For both the dissolution and transfer (acceptor compartment only) experiments, 5 mL of sample was manually withdrawn from the dissolution vessel using a Fortuna® Optima® glass syringe (Poulten & Graf GmbH, Wertheim, Germany) which was attached to a stainless-steel cannula and filtered with a 10 µm filter (Erweka GmbH, Heusenstamm, Germany). Each sample was immediately filtered using a 0.45 µm Whatman® PTFE filter (Rezist 30, GE Healthcare UK Ltd., Buckinghamshire, UK), where the

PT

initial 4.5 mL of each sample was immediately returned to the vessel and to avoid subsequent precipitation from the solution, the remaining 500 µL appropriately diluted with a mixture of

AC

CE

PT E

D

MA

NU

SC

RI

acetonitrile and water (ratio 1:1).

12

ACCEPTED MANUSCRIPT 2.4.5. Assessment of the results from dissolution and transfer experiments Dissolution: The dissolution of Nizoral® in the gastric compartment was performed to evaluate the maximum fraction of KTZ dissolved during the transfer experiment whilst the dissolution in the intestinal compartment was undertaken to establish the initial dissolution rate (z-value) according to the

PT

method of Nicolaides et al. (Nicolaides et al., 2001).

RI

Transfer experiments:

SC

Based on the concentration profile of drug measured in the acceptor compartment during the experiment (i.e. representing the transfer profile), the lag time (representing the delay before

NU

appearance of drug in the acceptor compartment) was visually identified. The fraction of drug transferred from the stomach to the intestine as either dissolved (fd) or undissolved (fs) material, was calculated according to the method of Berlin et al. (Berlin et al., 2014). The maximum observed

MA

concentration in solution in the acceptor compartment prior to precipitation was defined as the critical supersaturation concentration (CSC). The degree of supersaturation (DOS) was calculated

𝐷𝑂𝑆 =

𝐶𝑆𝐶 𝐶𝑠

PT E

supersaturation concentration.

D

according equation (1), whereby Cs represents equilibrium solubility and CSC the critical

Equation (1)

CE

Precipitation was described using a first order precipitation rate constant (PRC) calculated according to equation (2). In this case, Pt is the amount of drug precipitated and Ct the amount of KTZ dissolved

𝑃𝑡 = 𝐶𝑡 𝑃𝑅𝐶

AC

at time t, in the intestine.

Equation (2)

Following the drop in drug concentration in the acceptor phase after precipitation, the equilibrium concentration i.e. final concentration (FC) is reached. The area under the curve (AUC) was calculated according to the trapezoidal rule. Further information on the qualitative and quantitative evaluation of the profile obtained from the transfer model is presented in Figure S1 (see Supplementary Information).

13

ACCEPTED MANUSCRIPT 2.5. HPLC analysis: Samples generated from the solubility, dissolution and transfer experiments were analyzed for KTZ concentration by HPLC using an endcapped (5µm) LiChroCart 150-4.6 RP-column (Merck KGaA, Darmstadt, Germany). All components of the HPLC system were part of the La ChromElite model from VWR Hitachi which consisted of an L-2400 UV detector, L-2300 column oven, L-2200 auto sampler and L-2130 pump. EZChrom Elite software version 3.3.2 SP2 was used for the evaluation of

PT

the chromatograms (VWR, Darmstadt, Germany). The mobile phase comprised of 47 % Acetonitrile and 53 % buffer (~0.015% trimethylamine and

RI

~0.075% phosphoric acid adjusted to a pH of 3.3). The flow rate was 2 mL/min and the injection

SC

volume was 20 µL. The temperature of the column was maintained at 55°C and detection wavelength was set at 256 nm. Under these conditions, KTZ typically eluted at 1.8 min. LOD and LOQ were found to be 0.04 µg/mL and 0.06 µg/mL, respectively. A calibration curve in the 10 – 50 µg/ml

MA

NU

concentration range was prepared fresh for each analytical run and were all linear with a R2 ≥ 0.9995.

2.6. Assessment of Pharmacokinetic data for Nizoral®

D

The plasma concentration profiles measured in humans following oral dosing of the solution and commercially available tablet formulations (each as 200 mg) were manually extracted from the mean

PT E

plasma profiles from Huang et al. (Huang et al., 1986). Each formulation was administered with 200 mL of water to 24 healthy, fasting, male volunteers in a cross-over study design.

CE

For Stella® PBPK modeling, the absorptive kinetics and post absorptive disposition kinetics are required in order to simulate the plasma concentrations. For this purpose, the distribution and elimination kinetics after intravenous (i.v.) injection are required. Unfortunately, i.v. data was not

AC

available in the literature for KTZ. Since relative bioavailability for the oral solution was expected to be closest to the absolute bioavailability (highest drug exposure among the published PK studies in comparison to the tablet formulation) the absorption and elimination kinetics and the volume of distribution were estimated using the mean plasma profile from the oral solution. Therefore, the mean plasma profile for the oral solution from Huang et al. was fitted using WinNonlin®, (WinNonlin® Professional v. 5.1, Pharsight Corporation, Mountain view, CA, USA) and the required absorptive and post absorptive parameters were estimated by the software. KTZ was found to follow a one compartmental model (R2=0.9997), where the absorption and elimination rate constants were estimated to be 1.93 h-1 and 0.45 h-1, respectively. The apparent 14

ACCEPTED MANUSCRIPT volume of distribution (Vd/F), where Vd is adjusted for bioavailability factor F, was calculated to be 26.5 L.

2.7. Physiologically-based Pharmacokinetic modelling: Plasma profiles of KTZ were simulated using Stella® software (v. 9.1.3, Isee Sytems, Inc., Lebanon,

PT

NH, USA). The in silico model to examine the interplay between dissolution, supersaturation, precipitation and absorption is based on a modification of the previously published model by Berlin

RI

et al. (Berlin et al., 2014). A graphical representation of the model is given in fig. 2.

SC

The stomach is assumed to be a non-absorptive compartment in the model and consists of a gastric volume of 250 mL (50 mL residual volume and 200 mL co-administered water volume). To simulate the plasma profile for the tablet, dissolution of the formulation and gastric emptying (GE) of the fluid

NU

volume are considered to take place simultaneously. The GE rates applied in the PBPK model are the transfer rates as applied in the transfer model (section 2.4.3).

MA

The fraction dissolved (fd) (equation (3)) and fraction solid (fs) (equation (4)) were calculated according to the method described by Berlin et al. (Berlin et al. 2014). Td and Ts reflect the transfer of

𝑑𝑇𝑑 𝑑𝑡

= 𝐷𝑘𝑒 𝑓𝑑

PT E

of KTZ used in the experiments.

D

dissolved and solid KTZ from the stomach to the intestine, respectively, whilst D relates to the dose

Equation (3)

𝑑𝑇𝑠 𝑑𝑡

= 𝐷𝑘𝑒 𝑓𝑠

Equation (4)

For the in silico simulations, the volumes used for the intestinal compartment were based on the

CE

volumes used in the transfer model (section 2.4.3.).

AC

For the modelling aspect, as soon as CSC in the simulation is exceeded, the precipitation of KTZ occurs using the first order precipitation rate constant (PRC) determined from the respective in vitro experimental setup. The parameters used from each experimental setup for the simulations are given in table 2.

The dissolution of undissolved drug and redissolution of the precipitate in the intestine was assumed to follow a modification of the Noyes-Whitney equation (Dressman and Reppas, 2000) and is expressed by the z-values as described in section 2.4.5. This approach has shown its utility in a number of previous PBPK studies (Berlin et al., 2014; Juenemann et al., 2011; Nicolaides et al., 2001; Shono et al., 2011; Wagner et al., 2012).

15

ACCEPTED MANUSCRIPT The absorption rate constant (k01) and elimination rate constant (k10), calculated as described in

MA

NU

SC

RI

PT

section 2.6., were 1.93 and 0.45 h-1, respectively.

PT E

D

Fig. 2. Overview of Stella® PBPK model to simulate the pharmacokinetic (PK) profile of KTZ

CE

2.8. Evaluation of in vitro and in silico data: In vitro Transfer data:

AC

All statistical comparisons were performed using SigmaPlot® for Windows Version 11.0 (Systat Software, Inc. 2008). For the comparison of two different groups, a two tailed t-test was utilized whilst for the comparison of more than two groups, an analysis of variance (ANOVA) was applied. When the normality test passed, an ANOVA with pairwise multiple comparisons using the Holm-Sidak method was used. When the normality test failed, a Kruskal-Wallis ANOVA on ranks with subsequent pairwise multiple comparisons using the Tukey test was applied. For each of the statistical methods, significance was proposed when the calculated p-value was found to be less than 0.05.

16

ACCEPTED MANUSCRIPT In vivo & in silico plasma profiles: The simulated profiles were compared to the plasma profiles reported in the literature (Huang et al., 1986) using the difference factor (f1), calculated according the method of MOORE et al. (MOORE et al., 1996). Additionally, the simulated and observed values for the different pharmacokinetic parameters were evaluated using point estimates ratios for Cmax [Cmax (simulated)/Cmax (observed)], Tmax [Tmax (simulated)/Tmax (observed)] and AUC [AUC (simulated)/AUC (observed)]. The simulated

RI

were in the range between 0.80 and 1.25 (Marston and Polli, 1997).

PT

versus observed profiles were deemed to be equivalent (i.e. bioequivalent) when the point estimates

SC

3. Results and Discussion 3.1. Solubility experiments:

NU

As anticipated for this weak base, the solubility of KTZ shows a high degree of pH dependency, with the solubility at pH 1 greater than 20 mg/ml and at pH 6.5, reducing significantly to 4.2 μg/ml.

MA

Additionally for this moderately lipophilic drug, bile salts have an impact on solubility with an approximate 5-fold elevation in solubility from 4.2 μg/ml in blank buffer at pH 6.5 to 21.9 µg/mL in

(see Supplementary Information)

D

FaSSIF-V1. A detailed summary of the solubility of KTZ in different media can be viewed in Table S2

PT E

To replicate the different experimental conditions evaluated, equilibrium solubility was measured in the media representing the acceptor compartment after complete transfer of the donor compartment. Solubility in the acceptor compartment was only minimally altered with solubility

CE

values ranging between 19.2 and 33.6 µg/mL, and the corresponding pH ranging between 5.8 and 6.0. Interestingly, these results seem to slightly under predict the equilibrium solubility measured in

AC

vivo from previous published data by Psachoulias et al. (Psachoulias et al., 2011). In this case, the median equilibrium solubility of KTZ from luminal samples collected from the upper small intestine were found to be 36 or 73 µg/mL after dosing of either 100 mg or 300 mg of KTZ, respectively. The corresponding mean solubility values were reported to be 449 and 404 µg/mL, respectively. The difference between in vitro and in vivo solubility may be explained by the inherent differences in GI pH values. Whilst the pH for each in vitro experiment varied in a negligible range and was always greater than 5.7, in the in vivo study, the pH of the luminal samples used to evaluate the equilibrium solubility varied more considerably and luminal samples with a pH greater than 3.6 were included for assessment of in vivo solubility. As a consequence, solubility was higher in the luminal samples collected. 17

ACCEPTED MANUSCRIPT Taking into consideration the volume of FaSSGF-V2 at pH 2.0 required to dissolve a single dose of one Nizoral® tablet (200mg), approximately 41 mL is required; whilst in FaSSIF-V2 at pH 6.5, 33 L of media is required. Based on this information, it is anticipated that there will be no solubility restrictions in the fasted stomach; however restrictions in the intestine will be anticipated.

PT

3.2. Dissolution experiments: The dissolution behavior of the Nizoral® tablets (200 mg) was evaluated in FaSSGF-V2 and FaSSIF-V2.

RI

In FaSSGF-V2, greater than 80 % dissolution was observed after 20 minutes and complete dissolution after 45 minutes, whilst in FaSSIF-V2 only 2 % of the dose went into solution over the same time

SC

period.

Based on the dissolution profile in FaSSIF-V2, the initial dissolution rate (z-value) for the Nizoral®

NU

tablet was calculated to be 0.1109 mL/mg2/3/h. This value was used to represent the dissolution of

MA

undissolved drug in the small intestine in the Stella® PBPK model.

3.3. Transfer experiments:

D

3.3.1. Impact of disintegration of the formulation

PT E

To make the transfer model more physiologically relevant and allow the impact of simultaneous disintegration, dissolution and gastric emptying to occur, the gastric contents were transferred directly after the formulation was placed into the donor compartment using the different transfer

CE

rates into the intestinal compartment.

To evaluate the impact of the updated experimental setup, the transfer profile of a predissolved

AC

Nizoral® tablet was compared to the transfer profile where concurrent disintegration and dissolution occurred. As can be seen in table 2 and fig. 3, differences can be observed. For the predissolved material, the time at which the maximum concentration is achieved prior to precipitation, occurs earlier. Additionally, PRC is significantly greater for the predissolved formulation compared to when disintegration of the tablet is allowed to occur. Given that this tablet is relatively rapidly dissolving (> 80% in FaSSGF-V2 in 30 min), the other parameters including CSC, AUC and DOS were not significantly different. However, it needs to be considered that for formulations of other weak basic drugs with different dissolution properties in the donor compartment, disintegration may have a more pronounced impact on the transfer profiles observed. Given that disintegration and dissolution

18

ACCEPTED MANUSCRIPT of the formulation influences supersaturation and precipitation, it is important that this is considered in the transfer model. The advantage of the transfer model compared to other in vitro approaches (e.g.”BioGit” model described by Kourentas et al.) (Kourentas et al., 2016) is that this model allows the evaluation of

PT

dissolution of the formulation on supersaturation and precipitation in the intestinal compartment.

3.3.2. Impact of gastrointestinal volume:

RI

Using the accepted residual volume in the fasting stomach (between 25 and 50 mL) (Goetze et al.,

SC

2009; Gruber et al., 1987; Mudie et al., 2014) and co-administration of 200 mL of water in a typical clinical trial, the volume in the donor compartment was set to 250 mL. Therefore, for all transfer experiments the starting volume of the gastric compartment was set to 250 mL using FaSSGF-V2 at a

NU

pH of 2.0.

Based on literature data, the volume of the acceptor compartment of the transfer model was

MA

reduced to 350 mL (from 500 mL in the original model) in order to better reflect the anticipated proximal intestinal. To investigate the impact of volume on the transfer profile, experiments were

D

performed using 100 mL, 350 mL and 500 mL of FaSSIF-V2 (Marciani et al., 2010; Schiller et al., 2005;

PT E

Sutton, 2009).

As shown in fig. 3 and table 2, the impact of intestinal volume on the transfer profiles of KTZ is given. Using the smallest intestinal volume (100 mL), the concentration in the acceptor compartment

CE

increased the fastest with the concentration measured after five minutes being approximately 5-fold greater compared to when 500 mL was used (133.5 versus 27.9 µg/mL, respectively). As a consequence, precipitation occurred sooner and faster compared to either the 350 mL or 500 mL

AC

intestinal volume, which both appeared to be almost similar. Statistically, intestinal volume had a significant effect on CSC, DOS and PRC, especially between 100 and 350/500 mL, but had no effect on the AUC of the resulting transfer profiles. By evaluating PRC, an increase in volume leads to a significant decrease in precipitation velocity. A positive correlation between DOS and PRC was observed, whereby the highest DOS resulted in the fastest PRC, and inversely the lowest DOS demonstrated

the

slowest

PRC

(R2=0.93)

(Figure

S3

in

Supplementary

Information).

19

ACCEPTED MANUSCRIPT 3.3.3. Impact of transfer rate: Under fasting conditions, gastric emptying following a glass of water is known to occur by a first order kinetic (Hunt, 1956; Moore et al., 1984; Steingoetter et al., 2006). Based on literature data to reflect the average fasting gastric emptying kinetics, a first order transfer rate with a half-life of 9 min was proposed in the standard transfer model setup (Adkin et al., 1995; Murray et al., 1994; Oberle et al., 1990; Steingoetter et al., 2006; Wilding et al., 1994). However, to cover the range of gastric

PT

emptying patterns described in the literature, several first and zero order transfer rates were applied. To simulate a very slow gastric emptying rate, as can be anticipated during phase one of the

RI

Migrating Motor Complex (MMC), a 22.4 min first order half-life was used (Oberle et al., 1990). In order to examine the influence of a very rapid gastric emptying, such as can be anticipated during

SC

MMC phase three; the formulation was predissolved in the gastric compartment and poured directly into the acceptor compartment (i.e. “dumping”). Experiments were also conducted using the zero

NU

order transfer rates of 2 mL/min and a 9 mL/min flow rate as used in the original transfer model (Kostewicz et al., 2004).

MA

As can be seen in fig. 3 and table 2, the transfer rate has an enormous impact on the resulting transfer profiles for KTZ. For the different transfer rates, CSC varies between 150 µg/mL and 300 µg/mL and the time of occurrence of precipitation (Tmax) ranged between 5 min and 60 min. As

D

expected, the fastest gastric emptying rate (e.g. dumping) using predissolved drug, revealed the

PT E

overall highest CSC and DOS values. However, there seems to be an upper limit for the impact of transfer rate on precipitation. At a transfer rate of 9 mL/min, PRC was of 2.3 h-1 and at the fastest transfer as in the “dumping” method, PRC was significantly lower at 1.8 h-1.

CE

In general, an increase in transfer rate resulted in a higher CSC, DOS and faster PRC and an earlier onset of precipitation (Tmax) but AUC was not altered for both the zero and first order transfer rates.

AC

The influence of disintegration on supersaturation and precipitation needs to be further evaluated as a faster gastric emptying rate does not necessarily mean a higher CSC and DOS. By applying a faster gastric emptying rate, the volume of media remaining in the gastric compartment is lower, which may reduce the disintegration and dissolution rate of the formulation in the gastric compartment and as a consequence result in lower CSC and DOS values. However, to eliminate the effect of disintegration and dissolution, experiments using 300 mg of pre-dissolved KTZ were performed. These experiments confirm the positive relation between transfer rate and CSC/DOS (Ruff et al., 2013)

20

ACCEPTED MANUSCRIPT 3.3.4. Influence of bile salt concentration: For the updated version of the transfer model, FaSSIF-V2 rather than FaSSIF-V1 was taken as the standard medium to simulate the contents of the small intestine (Jantratid et al., 2008). Given the effect of bile salts on solubility (section 3.1), experiments using relevant bile salt concentrations were conducted to evaluate the impact on the supersaturation and precipitation behaviour of KTZ. Therefore, four different experimental setups using the standard approach were conducted using: (1)

PT

compendial media (i.e. lacking bile salts) in both the donor and acceptor compartment, (2) standard transfer experiment using FaSSIF-V2 where dilution of the bile salts in the acceptor compartment

RI

occurs, (3) using FaSSIF-V1 also where dilution of bile salts in the acceptor compartment occurs, and (4) bile salt concentration in the intestinal compartment was maintained throughout the experiment

SC

which was achieved by having a similar bile salt concentration in the gastric compartment. Due to the high pH dependent solubility and fast dissolution of KTZ at pH 2.0, it was assumed that the increase

NU

in bile salts in the donor compartment is not likely to have a significant effect on the dissolution behaviour of Nizoral®.

MA

As anticipated, bile salts have a dramatic effect on the transfer profile (fig. 3). Whilst during the first 10 min, all four profiles were similar, thereafter the concentration of KTZ in the intestinal compartment using compendial media was lowest. In the case where a low concentration of bile salt

D

is present (i.e. compendial media and FaSSIF-V2), an earlier onset of precipitation (Tmax) and lower

PT E

CSC was observed when compared to the other experimental setups using either a higher bile salt concentration (FaSSIF-V1) or when the bile salt concentration is not diluted in the acceptor compartment during the experiment.

CE

To distinguish between the impact of bile salts on solubility and supersaturation/precipitation, a linear regression performed between PRC and DOS indicated a strong relationship between DOS and

AC

PRC (R2=0.95) which suggests that PRC is highly dependent upon the DOS (Figure S4 in Supplementary Information).

3.3.5. Impact of hydrodynamic conditions: To evaluate the influence of hydrodynamic conditions, experiments were conducted using a slow (50 rpm), intermediate (100 rpm) and fast (150 rpm) paddle speed. To reduce the risk for coning and to ensure that the dissolution from capsule formulations will occur, the agitation speed in the standard setup was 100 rpm.

21

ACCEPTED MANUSCRIPT Transfer experiments were conducted by placing the tablet directly in the donor compartment and in this case, the supersaturation and precipitation behaviour was a function of the dissolution of the tablet in the gastric compartment. For these experiments, the maximum amount of drug dissolved in the intestinal compartment was increased with faster paddle speed. Over the 50 to 150 rpm range, hydrodynamics had a significant effect on CSC, AUC, PRC and DOS. Over this range, CSC increased from 148.1 µg/mL to 264.5 µg/mL and DOS from 6.1 to 10.8 (table 2). The PRC also increased significantly with increasing paddle speed from 0.15 h-1 (50 rpm) to 1.9 h-1 (150 rpm). The onset of

PT

precipitation was quicker, whereby Tmax decreased from approximately 1 h to approximately 20 min, following an increase in paddle speed from 50 to 150 rpm. Although paddle speed had a significant

RI

effect on AUC, no linear correlation between hydrodynamics and AUC could be established.

SC

To isolate the influence of paddle speed on supersaturation and precipitation (and not dissolution), experiments were conducted using predissolved KTZ in the donor compartment. These experiments

NU

indicated a significant impact of agitation on the transfer profile. In general, precipitation occurred earlier and PRC increased with increasing agitation speed, which also had a significant influence of

MA

reducing the amount of drug in solution (i.e. AUC).

By taking the results for the oral solution and when disintegration of the tablet formulation was taken into account, it was found that hydrodynamics not only affect the dissolution characteristics of

D

the Nizoral® tablet formulation and therefore CSC, but the precipitation characteristics also. The

PT E

faster the paddle speed, the transfer profiles become more similar to the oral solution (earlier Tmax and higher CSC) and the faster the observed PRC. By assuming that the amount of drug available for absorption is represented by AUC, the importance

CE

of the interplay between the CSC and PRC needs to be evaluated. Since AUC is influenced by Tmax, CSC and PRC, it has to be considered that a low paddle speed results in a slower dissolution of the

AC

formulation in the donor compartment which consequently leads to a delayed increase in concentration (and a low CSC) in the acceptor compartment. As a consequence, a lower CSC resulted in a slower PRC.

22

ACCEPTED MANUSCRIPT Table 2: Influence of set-up parameters on supersaturation and precipitation characteristics of the Nizoral® tablet / oral solution (predissolved tablet).

Transfer rate Standard set-up (t½=9 min) a e

188.8 ± 5.5 * 207.8 ± 12.1 d 165.5 ± 9.8 c 188.8 ± 5.5

17935 ± 591 16456 ± 1057 17935 ± 591 * 12343 ± 1046 13646 ± 1836 17935 ± 591

a,f,h

PRC [h-1]

DOS

* 1.75 ± 0.1 2.08 ± 0.1 a

7.7 ± 0.2

PT

188.8 ± 5.5 209.4 ± 15.5

SC

Disintegration of formulation Standard set-up (disintegration and dissolution) a Predissolved b Intestinal volume Standard set-up (350 mL) a 100 mL c 500 mL d

AUC [min*µg*mL-1]

7.7 ± 0.2 * 9.5 ± 0.6 a 8.8± 0.5

1.75 ± 0.1 0.72 ± 0.0 a,f,g,h * 2.31 ± 0.1 a,e,g,h 1.37 ± 0.1 a,e,f,h 1.84 ± 0.0 e,f,g

7.7 ± 0.2 5.6 ± 0.3 a,f,h * 9.3 ± 0.4 a,e,g,h 5.2 ± 0.0 a,f,h 12.2 ± 0.2 a,e,f,g

1.75 ± 0.1 1.93 ± 0.1 j * 1.53 ± 0.1 i,k 1.98 ± 0.1 j

7.7 ± 0.2 8.6 ± 1.0 * 7.1 ± 0.2 k 9.1 ± 0.2 j

137.4 ± 7.4 * 227.2 ± 10.3 a,e,g,h 126.3 ± 1.2 a,f,h 301.7 ± 1.7 a,e,f,g

Bile salts Standard set-up (FaSSIF-V2) a Compendial i FaSSIF-V1 j FaSSGF-V2(HBS) k

188.8 ± 5.5 181.4 ± 5.7 j,k * 239.3 ± 6.6 a,i 230.7 ± 4,8 a,i

Paddle speed (simultaneously) Standard set-up (100 rpm) a 50 rpm l 150 rpm m

188.8 ± 5.5 * 148.1 ± 6.2 *,a,m 264.5 ± 9.9 *,a,l

17935 ± 591 * 29663 ± 302 a,m 21039 ± 2003 a,l

1.75 ± 0.1 * 0.15 ± 0.0 a,m 1.94 ± 0.1 l

7.7 ± 0.2 * 6.1 ± 0.3 a,m 10.8 ± 0.4 a,l

239.1 ± 4.8 209.4 ± 15.5 247.5 ± 22.5

34123 ± 3590 o,p * 16457 ± 1057 n 17387 ± 3143 n

0.61 ± 0.2 o,p * 2.08 ± 0.1 n 2.16 ± 0.2 n

9.8 ± 0.2 8.5 ± 0.7 10.5 ± 0.9

NU

Zero order: 2 mL/min Zero order: 9 mL/min f First order: t½=22.4 min g Gastric dumping h

PT E

D

MA

17935 ± 591 12612 ± 33 j 22454 ± 1561 a,i 17720 ± 1165

AC

CE

Paddle speed (predissolved) 50 rpm n 100 rpm o 150 rpm p

16488 ± 1048 * 14276 ± 250 13021 ± 177 h 18820 ± 221 g

8.5 ± 0.7

1.75 ± 0.1 * 3.17 ± 0.0 a,d 2.12 ± 0.4 c

RI

CSC [µg/mL]

* ANOVA showed significant impact on parameter for the respective setup parameter, superscript letters indicate statistically significant difference from the astandard set-up (disintegration and dissolution of formulation, 350 mL of intestinal volume, first order transfer rate t½=9 min, FaSSIF-V2 in intestinal compartment, 100 rpm agitation speed), bpredissolved formulation, c100 mL of intestinal volume, d500 mL of intestinal volume, ezero order transfer rate: 2 mL/min, fzero order transfer rate: 9 mL/min, gfirst order transfer rate: t½=22.4 min, hgastric dumping as transfer rate, icompendial media, j FaSSIF-V1 as media, kFaSSGF-V2(HBS) as media, l50 rpm agitation speed (simultaneous transfer), m 150 rpm agitation speed (simultaneous transfer), n50 rpm agitation speed (transfer with predissolved material), o100 rpm agitation speed (transfer with predissolved material), p150 rpm agitation speed (transfer with predissolved material)

23

ACCEPTED MANUSCRIPT 3.3.6. Summary of the in vitro results: All of the parameters investigated in the transfer model had an impact, to varying degrees, on the supersaturation and precipitation behaviour of KTZ. Based on these results, hydrodynamics showed a significant impact not only on its importance on disintegration and dissolution, but as shown by the results from the oral solution, on supersaturation and precipitation also. The greater the paddle speed, dissolution is accelerated and as a consequence higher concentrations are achieved in the

PT

intestinal compartment (which may consequently drive faster precipitation) and when evaluated in isolation, a higher paddle speed leads to a faster precipitation rate. Another equally important

RI

parameter influencing the supersaturation and precipitation behaviour is transfer rate. Generally, the faster the gastric emptying rate, the greater the degree of supersaturation, however any benefits

SC

achieved by a higher concentration, are offset by a faster precipitation rate. The influence of intestinal volume and the presence of bile salts, whilst had a significant influence, their effect was

NU

less pronounced than the other parameters evaluated, suggesting that slight variations in the volume and media composition will not influence the results significantly.

MA

For most of the different transfer model parameters evaluated, a positive correlation between DOS and PRC could be identified, in this case the higher the DOS the faster the PRC. Interestingly, AUC which provides a measure for the amount of drug available for absorption, was generally less

D

sensitive on the transfer model parameters than CSC, PRC and DOS, suggesting that the amount of

AC

CE

PT E

drug in solution is a composite of numerous potentially competing parameters.

24

PT E

D

MA

NU

SC

RI

PT

ACCEPTED MANUSCRIPT

Fig. 3. The influence of: a) disintegration of the formulation, b) intestinal volume, c) transfer rate, d)

CE

bile salts, e) hydrodynamics for tablet formulation, f) hydrodynamics for an oral solution, on the transfer profile for KTZ.

AC

# theoretical concentration profile is not given in (fig. b and c) since each intestinal volume and flow rate exhibits a different theoretical concentration profile

25

ACCEPTED MANUSCRIPT 3.4. Simulation of plasma profiles (PBPK approach) In order to evaluate the suitability of the transfer model setup, the input parameters from each of the different experimental conditions were included in the Stella® model to simulate the plasma profile in humans. To improve the physiological relevance, a number of important changes were included in the original Stella® model published (Berlin et al., 2014). One important update to the model was that it now included a lag time for the disintegration of the formulation (as occurring in

PT

the donor compartment) as anticipated to occur in vivo. During this period in the simulation, only dissolved drug is transferred into the intestinal compartment. Another important update to the

RI

model is that the onset of precipitation only occurs when the simulated concentration exceeds the

SC

CSC.

Previous studies have shown that absorption has to be considered to accurately predict the precipitation behaviour in vivo (Carlert et al., 2010; Cristofoletti et al., 2016; Psachoulias et al., 2011;

NU

Taupitz et al., 2013). In the present study, the Stella® PBPK model was constructed to integrate the impact of absorption on the measured luminal concentrations and therefore to evaluate the impact

MA

of precipitation observed in vitro. Based on the different transfer model setups, the values for fraction dissolved (fd), PRC, CSC and final concentration (FC), as summarized in table 3, were incorporated directly in the PBPK model. All simulated profiles were directly compared to the mean

AC

CE

PT E

D

observed in vivo profile (Huang et al., 1986).

26

ACCEPTED MANUSCRIPT Table 3: Summary of in vitro transfer model parameters used for the in silico Stella® PBPK modelling. fd Setup variation

During

After

disintegration

disintegration

0.18

0.81

Standard setup

PRC

CSC

FC

[h-1]

[µg/mL]

[µg/mL]

1.75

188.8

35.8

2.08

209.4

46.5

1.37

Disintegration

PT

0.87*

35.4

RI

Predissolved

1.84

301.7

46.7

0.72

137.4

40.3

2.31

227.2

29.3

0.85

1.93

181.4

23.7

1.00

1.53

239.3

54.9

1.00

1.98

230.7

37.7

0.40

3.17

207.8

26.1

0.92

2.12

165.5

32.5

0.10

0.74

0.15

148.1

95.5

0.74

0.97

1.94

264.5

52.5

0.75*

0.61

239.1

84.6

0.87*

2.08

209.4

46.5

1.00*

2.16

247.5

47.0

Transfer rate First order (t ½ 22.4 min)

0.36 0.89*

Zero order (2 mL/min)

0.23

0.81

9 mL/min

0.22

0.65

0.24

FaSSIF-V1 (diluted)

0.22

FaSSIF-V2 (constant)

0.24

Intestinal volume 0.31

500 mL

0.19

PT E

D

100 mL

MA

Compendial media

NU

Bile salts

SC

Gastric dumping

0.58

126.3

Paddle speed (simultaneously) 50 rpm

CE

150 rpm

50 rpm 100 rpm 150 rpm

AC

Paddle speed (predissolved)

* Since drug is already predissolved no lag time for disintegration was included

3.4.1. Evaluation of the transfer model setup to accurately simulate the plasma profile: To evaluate the suitability of the updated Stella® model, plasma profile simulations of the oral solution were performed using in vitro data from transfer experiments using predissolved Nizoral® tablets. In the case of an oral solution, disintegration does not need to be taken into account, and in 27

ACCEPTED MANUSCRIPT this case, gastric emptying is the most relevant step for transporting the dissolved drug to the site of absorption. A detailed description and summary of the results for the oral solution is given in the supplementary information (table S3 and figure S5). To investigate the experimental conditions which best simulate the plasma profile using the updated Stella model, the in vitro data collected from the different transfer model setups was incorporated in the model.

PT

Impact of disintegration of the formulation on simulated plasma profiles

RI

The ability to accurately predict the plasma profiles was significant greater when disintegration and dissolution of the formulation in the gastric compartment was taken into account in the simulations

SC

(fig. 4). The point estimate ratios for Cmax (1.09) and AUC (0.94) were in the 90% confidence interval for the Cmax and AUC ratio values and the f1-value (14.23) also indicated similarity of the profiles.

NU

When disintegration of the formulation was not considered, an overprediction of the in vivo profile was observed. The point estimate ratios for Tmax and Cmax with values of 0.65 and 1.41, respectively, and the f1-test (34.69) failed the requirements to verify similarity (table 4). Based on these results,

MA

the inclusion of disintegration and dissolution data is required for the accurate prediction of the plasma profiles by the PBPK model.

D

Impact of intestinal volume on simulated plasma profiles

PT E

Based on the in vitro data from the transfer model using intestinal volumes in the 100 mL to 500 mL range, a significantly better fit with the in vivo plasma profile was achieved using a 350 mL intestinal compartment volume. In this case, the point estimate ratios for Cmax and AUC were in the requested

CE

range for bioequivalence (fig. 4 and table 4). The f1 factor also indicated a greater similarity with the actual plasma profile when using 350 mL (14.2) compared to either the 100 mL (31.5) or 500 mL

AC

(25.5) volumes. This is consistent with the current information that intestinal volume is lower than originally thought (Schiller et al., 2005) such that the 350 mL is suitable for accurately predicting the plasma profile.

Impact of transfer rate on the simulated plasma profiles As shown in table 4 and fig. 4, the in silico simulations using an in vitro first order transfer rate with a half-life time of 9 minutes gave the best prediction, indicating the need for using a physiologically relevant gastric emptying rate in the transfer model. This rate represents an intermediate emptying pattern that can be anticipated in the MMC in healthy human volunteers. Using the slowest first order transfer rate (half-life time 22.4 min), Tmax, Cmax, and AUC were all underpredicted with an unsatisfactory f1-value of 31.9. Both of the zero order transfer rates investigated (2 and 9 mL/min) 28

ACCEPTED MANUSCRIPT did not provide a good simulation of the plasma profile. Whilst the point estimate ratios were found to be reasonable using the 2 mL/min flow rate, the f1-value (32.96) reflects a poor prediction for the overall plasma profile. By applying the in vitro results from the “dumping” experiment, a clear over prediction of the profile was observed.

PT

Impact of bile salts on simulated plasma profiles As illustrated in table 4 and fig. 4, bile salt concentration in the transfer experiment had an impact on

RI

the resulting simulated plasma profiles. In the case of using FaSSIF-V2 (as in the standard setup), the mean plasma profile was simulated best. As mentioned in section 4.3.4, the in vitro profiles using

SC

FaSSIF-V2 or compendial media did not show any significant difference. As a result, the simulated Cmax and AUC values were in the 90% confidence interval of the ratio of AUC and Cmax values,

NU

nevertheless, based on the f1 factor, the value was less than 15 when using FaSSIF-V2 suggesting a greater similarity for the simulated profile for this media. In the case where the bile salt concentration was higher (FaSSIF-V1) or where bile salt concentration remained constant in the

MA

intestinal compartment over time (FaSSGF-V2(CBS)), the simulated plasma profiles were overpredicted, suggesting that the higher concentrations achieved in the intestinal lumen ,as

D

expected, result in the predicted plasma concentrations being higher. Therefore, it is proposed that

PT E

FaSSIF-V2 should be used as the media in the transfer model to reflect the GI luminal conditions.

CE

Impact of hydrodynamics on simulated plasma profiles: To evaluate the influence of hydrodynamics used in the transfer model on the simulation of the plasma profile using the Stella® model, the in vitro data collected at 50, 100 and 150 rpm using either

AC

the predissolved tablet to reflect “solution” or tablet formulation were used for the simulations (Huang et al., 1986) (Fig. 4 and table 4). Given the influence of hydrodynamics on disintegration, dissolution and subsequent supersaturation observed in the in vitro studies, this also had a significant impact on the resulting simulated plasma profiles. Utilizing a fast paddle speed (e.g. 150 rpm) led to a slight overprediction of the mean in vivo plasma profile, whereby at both 50 rpm and 100 rpm, the simulated profiles were closer to the observed plasma profile. The f1-value established for 100 rpm (14.2) indicated the greatest similarity to the in vivo profile compared to 50 rpm (16.6) and 150 rpm (36.5). These simulations suggest the suitability (in the case of Nizoral®) for using 100 rpm in the transfer model. 29

ACCEPTED MANUSCRIPT In the case of the oral solution (e.g. predissolved tablet) where paddle speed had no influence on the initial part of the transfer profile (in contrast to the tablet), the high absorption rate of KTZ (k01=1.93h-1) led to a complete absorption of the solubilized drug for all agitation speeds applied. Whilst there were significant differences in the in vitro precipitation behaviour as a function of paddle speed (section 3.3.5), the predicted plasma profiles using the different paddles speeds were all similar as CSC in simulations was not exceeded, and as a consequence precipitation did not occur (fig. 4). The PK parameters were in the required range for bioequivalence values and all f1-values

AC

CE

PT E

D

MA

NU

SC

RI

PT

were below 15 (table 4).

Fig. 4. Impact of a) disintegration and dissolution of the formulation, b) transfer rate, c) bile salt concentration and composition, d) intestinal volume, e) hydrodynamics using a Nizoral® tablet formulation, f) hydrodynamics using an oral solution on the simulated KTZ plasma profile using the Stella® model.

30

ACCEPTED MANUSCRIPT Table 4: Comparison of the observed and simulated plasma profiles using the f1-difference factor and point estimates ratios for Tmax, Cmax and AUC f1-test

Observed

-

Tmax

Cmax

AUC0-12h

[h]

[µg/mL]

[min*µg*mL-1]

2.0

3.26

13.4

pred.

ratio

pred.

ratio.

pred.

ratio

14.23

1.4

0.70

3.56

1.09

12.6

0.94

predissolved

34.69

1.3

0.65

4.60

1.41

16.5

1.23

Zero order (2 mL/min)

33.0

2.5

Zero order (9 mL/min)

29.5

First order (t1/2 = 22.4 min)

31.9

First order (t1/2 = 9.0 min)

14.2

Gastric dumping

33.5

RI

Non predissolved

Transfer Rate

3.6

1.10

14.0

1.04

1.3

0.65

4.6

1.41

16.0

1.19

1.5

0.75

2.4

0.74

8.9

0.66

1.4

0.70

3.6

1.09

12.6

0.94

0.8

0.40

3.3

1.00

10.7

0.80

15.5

1.4

0.70

3.8

1.18

13.6

1.01

14.2

1.4

0.70

3.6

1.09

12.6

0.94

38.8

1.4

0.70

4.9

1.52

17.5

1.31

38.8

1.4

0.70

4.9

1.52

17.5

1.31

31.5

1.3

0.65

2.4

0.74

8.5

0.63

14.2

1.4

0.70

3.6

1.09

12.6

0.94

22.5

1.4

0.70

4.3

1.33

15.4

1.15

50 rpm

16.6

1.4

0.70

3.1

0.94

11.0

0.82

100 rpm

14.2

1.4

0.70

3.6

1.09

12.6

0.94

150 rpm

36.5

1.3

0.65

4.8

1.47

17.1

1.28

NU

MA

Compendial media

D

FaSSIF-V2 (diluted)

FaSSIF-V2 (constant) Intestinal Volume

350 mL 500 mL

CE

100 mL

PT E

FaSSIF-V1 (diluted)

SC

1.25

Bile Salts

Hydrodynamics (tablet)

AC

Nizoral® tablet formulation

PT

Disintegration of the Formulation

Observed

-

1.0

5.16

18.2

Solution

Hydrodynamics (oral solution) 50 rpm

14.77

1.4

1.4

4.58

0.89

16.5

0.91

100 rpm

14.35

1.3

1.3

4.65

0.90

16.5

0.91

150 rpm

11.07

1.2

1.2

4.7

0.91

16.7

0.92

31

ACCEPTED MANUSCRIPT 3.4.2. Interplay of permeability on supersaturation and precipitation: Permeability & critical supersaturation concentration (CSC) As mentioned previously, the in vitro transfer model does not take absorption into account. To get a better understanding of the influence that a drugs’ permeability might have on the simulated profiles based on the supersaturation and precipitation observed in vitro, a parameter sensitivity analysis of absorption rate constant (k01) was performed. As shown in fig. 5, the results from the simulations

PT

confirm the assumption that a high absorption can reduce the likelihood of precipitation and that a low absorption may increase the risk for precipitation occuring in vivo. The biggest impact of

RI

precipitation on the simulations was observed when using the lowest permeability (k01), as shown by

SC

the lowest predicted plasma concentrations (fig. 5). A k01 of 1.2 h-1 was found to be the theoretical “limit” for precipitation, above which the simulated luminal concentrations did not exceed CSC and consequently precipitation did not occur in the luminal compartment. A parameter sensitivity

NU

analysis performed on CSC resulted in a similar effect. By decreasing the CSC from 210 µg/mL to 180 µg/mL and by applying a PRC of 1.75 h-1 (as from the experiments) the simulated luminal

MA

concentrations exceeded the CSC and precipitation in the intestinal compartment was initiated resulting in lower simulated plasma concentrations. Also, decreasing CSC to 10 µg/mL, did not increase the degree of precipitation greatly and therefore having no dramatic impact on the

D

simulated plasma concentration time profile. Therefore, these results illustrate the impact of the

PT E

interplay between CSC and absorption on the resulting plasma concentration time profiles and the

CE

importance to couple the in vitro data from the transfer model using a suitable in silico approach.

Precipitation rate constant (PRC)

AC

Parameter sensitivity analysis of PRC was also performed to investigate its influence on the resulting simulated plasma profile. To ensure that precipitation would occur in the simulations, the CSC was reduced to 180 µg/mL and the influence of PRC in the range between 0 and 5 h-1 was investigated. When reducing CSC slightly and utilizing the observed in vitro PRC of 1.75 h-1, the plasma profile was significantly under predicted. When either negligible or no precipitation was assumed, the simulated profiles were closest to the in vivo profile (Fig. 5). These results are in alignment with previous work of Cristofeletti et al. (Cristofoletti et al., 2016) where they tried to simulate KTZ plasma profiles using a PBPK approach (SimCyp). For their simulations they had to assume negligible precipitation using a slow PRC of 0.01 h-1 in order to accurately predict the KTZ plasma profile. 32

CE

PT E

D

MA

NU

SC

RI

PT

ACCEPTED MANUSCRIPT

Fig. 5. Sensitivity analysis of a) absorption rate constant (k01), b) critical supersaturation

AC

concentration (CSC), c) precipitation rate constant (PRC) on the resulting simulated KTZ plasma concentration profile

#CSC was reduced to 180 µg/mL to enable precipitation in the in silico approach

3.4.3. Summary of in silico simulations The updated Stella model was found to be a suitable model to accurately incorporate the supersaturation and precipitation data from the in vitro transfer experiments and to evaluate their impact on the absorption of KTZ. Given the fast absorption rate of KTZ, solubilized drug in the 33

ACCEPTED MANUSCRIPT intestine was very quickly absorbed in the simulations, such that CSC could not be exceeded and as a consequence, precipitation that was observed in vitro did not occur in vivo. Interestingly, by performing a parameter sensitivity analysis of permeability at a fixed PRC, when using a lower absorption rate precipitation had a significant effect on the simulated profiles. For the KTZ example, disintegration and dissolution of the formulation, transfer rate, intestinal volume and hydrodynamics all had an impact on the simulated profile. Importantly, in the case of

PT

using data collected from the proposed standard transfer model, the simulated profile was closest to the mean observed in vivo plasma profile, thereby supporting the use of these optimized

RI

experimental conditions for the evaluation of supersaturation and precipitation.

SC

These results are consistent with previous approaches to investigate supersaturation and precipitation such as the gastrointestinal simulator (GIS) (Takeuchi et al., 2014), the biorelevant gastrointestinal transfer system (BioGit) (Kourentas et al., 2016), the TNO gastrointestinal model

NU

(TIM) (Blanquet et al., 2004), and the artificial stomach-duodenum model (ASD-model) (Mitra and Fadda, 2014), where more predictive results have been obtained when the experimental conditions

MA

were optimised to better reflect the in vivo GI physiology (Gao et al., 2010; Matsui et al., 2015,

D

2016)).

PT E

4. CONCLUSION:

The present study demonstrated the importance of developing a reliable tool to investigate the impact of supersaturation and precipitation on the bioavailability of weakly basic drugs. By taking the

CE

“average” physiological gastrointestinal conditions into consideration, a standard transfer model was proposed and the impact of physiological relevant extreme conditions on the supersaturation and

AC

precipitation behaviour of KTZ was investigated. Due to the lack of an absorptive compartment, the in vitro model overpredicted the precipitation behaviour of KTZ and indicates the need to take this into consideration in order to forecast its impact on absorption in vivo. Given the different absorption characteristics of drugs, the requirement to use an in vitro absorptive sink was considered to be inappropriate since in this case the experimental setup would need to be adjusted and validated for each absorption rate (low to high). Therefore, the in vitro model was successfully coupled with an updated Stella® in silico model and the impact of absorption on supersaturation and precipitation could be investigated. Consequently, the proposed standard transfer model gave the most accurate simulation of the observed profile and supports the assumption to make the in vitro model more physiologically relevant when coupling to PBPK 34

ACCEPTED MANUSCRIPT modelling. It has to be considered that for BCS class 2 compounds like KTZ, the impact of the precipitation observed in vitro may either be negligible or non-existent in vivo, whereby for BCS class IV compounds, precipitation may significantly reduce the amount of drug available for absorption. Nevertheless, further investigations using additional weakly basic drugs, such as those belonging to BCS IV, are needed in order to explore the complexity of the impact of absorption on supersaturation

PT

and precipitation further.

RI

Acknowledgments:

The authors would like to thank Nadine Banna for her support with some of the practical work

SC

undertaken (Intern at the Institute of Pharmaceutical Technology, Goethe University) and Dr. Mark Berlin for his assistance for the update of the Stella® model. These studies were funded by the

NU

Innovative Medicines Initiative Joint Undertaking under grant agreement no. 115369. The resources consist of financial support from the European Union’s Seventh Framework Program (FP7/2007-

AC

CE

PT E

D

MA

2013) and the kind contributions of the EFPIA partners.

35

ACCEPTED MANUSCRIPT References Adkin, D.A., Davis, S.S., Sparrow, R.A., Huckle, P.D., Phillips, A.J., Wilding, I.R., 1995. The effects of pharmaceutical excipients on small intestinal transit. Br J Clin Pharmacol 39, 381-387. Annaert, P., Brouwers, J., Bijnens, A., Lammert, F., Tack, J., Augustijns, P., 2010. Ex vivo permeability experiments in excised rat intestinal tissue and in vitro solubility measurements in aspirated human

PT

intestinal fluids support age-dependent oral drug absorption. Eur J Pharm Sci 39, 15-22.

RI

Berlin, M., Przyklenk, K.H., Richtberg, A., Baumann, W., Dressman, J.B., 2014. Prediction of oral

SC

absorption of cinnarizine--a highly supersaturating poorly soluble weak base with borderline permeability. Eur J Pharm Biopharm 88, 795-806.

NU

Bevernage, J., Brouwers, J., Clarysse, S., Vertzoni, M., Tack, J., Annaert, P., Augustijns, P., 2010. Drug supersaturation in simulated and human intestinal fluids representing different nutritional states. J

MA

Pharm Sci 99, 4525-4534.

Blanquet, S., Zeijdner, E., Beyssac, E., Meunier, J.P., Denis, S., Havenaar, R., Alric, M., 2004. A dynamic

D

artificial gastrointestinal system for studying the behavior of orally administered drug dosage forms

PT E

under various physiological conditions. Pharm Res 21, 585-591. Blum, R.A., D'Andrea, D.T., Florentino, B.M., Wilton, J.H., Hilligoss, D.M., Gardner, M.J., Henry, E.B.,

CE

Goldstein, H., Schentag, J.J., 1991. Increased gastric pH and the bioavailability of fluconazole and ketoconazole. Ann Intern Med 114, 755-757.

AC

Brouwers, J., Brewster, M.E., Augustijns, P., 2009. Supersaturating drug delivery systems: the answer to solubility-limited oral bioavailability? J Pharm Sci 98, 2549-2572. Brouwers, J., Tack, J., Lammert, F., Augustijns, P., 2006. Intraluminal drug and formulation behavior and integration in in vitro permeability estimation: a case study with amprenavir. J Pharm Sci 95, 372-383. Carlert, S., Palsson, A., Hanisch, G., von Corswant, C., Nilsson, C., Lindfors, L., Lennernas, H., Abrahamsson, B., 2010. Predicting Intestinal Precipitation-A Case Example for a Basic BCS Class II Drug. Pharm Res-Dordr 27, 2119-2130. 36

ACCEPTED MANUSCRIPT Clarysse, S., Psachoulias, D., Brouwers, J., Tack, J., Annaert, P., Duchateau, G., Reppas, C., Augustijns, P., 2009a. Postprandial changes in solubilizing capacity of human intestinal fluids for BCS class II drugs. Pharm Res 26, 1456-1466. Clarysse, S., Tack, J., Lammert, F., Duchateau, G., Reppas, C., Augustijns, P., 2009b. Postprandial evolution in composition and characteristics of human duodenal fluids in different nutritional states.

PT

J Pharm Sci 98, 1177-1192. Cristofoletti, R., Patel, N., Dressman, J.B., 2016. Differences in Food Effects for 2 Weak Bases With

RI

Similar BCS Drug-Related Properties: What Is Happening in the Intestinal Lumen? J Pharm Sci.

SC

Dressman, J.B., Reppas, C., 2000. In vitro-in vivo correlations for lipophilic, poorly water-soluble

NU

drugs. Eur J Pharm Sci 11 Suppl 2, S73-80.

Fei, Y., Kostewicz, E.S., Sheu, M.T., Dressman, J.B., 2013. Analysis of the enhanced oral bioavailability

MA

of fenofibrate lipid formulations in fasted humans using an in vitro-in silico-in vivo approach. Eur J Pharm Biopharm 85, 1274-1284.

D

Gao, Y., Carr, R.A., Spence, J.K., Wang, W.W., Turner, T.M., Lipari, J.M., Miller, J.M., 2010. A pH-

PT E

dilution method for estimation of biorelevant drug solubility along the gastrointestinal tract: application to physiologically based pharmacokinetic modeling. Mol Pharm 7, 1516-1526. Ghasemi, J., Saaidpour, S., 2007. Quantitative structure-property relationship study of n-octanol-

AC

604, 99-106.

CE

water partition coefficients of some of diverse drugs using multiple linear regression. Anal Chim Acta

Glomme, A., Marz, J., Dressman, J.B., 2005. Comparison of a miniaturized shake-flask solubility method with automated potentiometric acid/base titrations and calculated solubilities. J Pharm Sci 94, 1-16. Goetze, O., Treier, R., Fox, M., Steingoetter, A., Fried, M., Boesiger, P., Schwizer, W., 2009. The effect of gastric secretion on gastric physiology and emptying in the fasted and fed state assessed by magnetic resonance imaging. Neurogastroenterol Motil 21, 725-e742.

37

ACCEPTED MANUSCRIPT Gruber, P., Rubinstein, A., Li, V.H., Bass, P., Robinson, J.R., 1987. Gastric emptying of nondigestible solids in the fasted dog. J Pharm Sci 76, 117-122. Huang, Y.C., Colaizzi, J.L., Bierman, R.H., Woestenborghs, R., Heykants, J., 1986. Pharmacokinetics and dose proportionality of ketoconazole in normal volunteers. Antimicrob Agents Chemother 30, 206-210.

PT

Hunt, J.N., 1956. Some properties of an alimentary osmoreceptor mechanism. J Physiol 132, 267-288. Jantratid, E., Janssen, N., Reppas, C., Dressman, J.B., 2008. Dissolution media simulating conditions in

RI

the proximal human gastrointestinal tract: an update. Pharm Res 25, 1663-1676.

SC

Juenemann, D., Jantratid, E., Wagner, C., Reppas, C., Vertzoni, M., Dressman, J.B., 2011. Biorelevant

NU

in vitro dissolution testing of products containing micronized or nanosized fenofibrate with a view to predicting plasma profiles. Eur J Pharm Biopharm 77, 257-264.

MA

Kostewicz, E.S., Abrahamsson, B., Brewster, M., Brouwers, J., Butler, J., Carlert, S., Dickinson, P.A., Dressman, J., Holm, R., Klein, S., Mann, J., McAllister, M., Minekus, M., Muenster, U., Mullertz, A.,

D

Verwei, M., Vertzoni, M., Weitschies, W., Augustijns, P., 2014. In vitro models for the prediction of in

PT E

vivo performance of oral dosage forms. Eur J Pharm Sci 57, 342-366. Kostewicz, E.S., Brauns, U., Becker, R., Dressman, J.B., 2002. Forecasting the oral absorption behavior

19, 345-349.

CE

of poorly soluble weak bases using solubility and dissolution studies in biorelevant media. Pharm Res

AC

Kostewicz, E.S., Wunderlich, M., Brauns, U., Becker, R., Bock, T., Dressman, J.B., 2004. Predicting the precipitation of poorly soluble weak bases upon entry in the small intestine. J Pharm Pharmacol 56, 43-51.

Kourentas, A., Vertzoni, M., Stavrinoudakis, N., Symillidis, A., Brouwers, J., Augustijns, P., Reppas, C., Symillides, M., 2016. An in vitro biorelevant gastrointestinal transfer (BioGIT) system for forecasting concentrations in the fasted upper small intestine: Design, implementation, and evaluation. Eur J Pharm Sci 82, 106-114.

38

ACCEPTED MANUSCRIPT Lobell, M., Hendrix, M., Hinzen, B., Keldenich, J., Meier, H., Schmeck, C., Schohe-Loop, R., Wunberg, T., Hillisch, A., 2006. In silico ADMET traffic lights as a tool for the prioritization of HTS hits. ChemMedChem 1, 1229-1236. Marciani, L., Cox, E.F., Hoad, C.L., Pritchard, S., Totman, J.J., Foley, S., Mistry, A., Evans, S., Gowland, P.A., Spiller, R.C., 2010. Postprandial changes in small bowel water content in healthy subjects and

PT

patients with irritable bowel syndrome. Gastroenterology 138, 469-477, 477 e461. Marston, S.A., Polli, J.E., 1997. Evaluation of direct curve comparison metrics applied to

RI

pharmacokinetic profiles and relative bioavailability and bioequivalence. Pharm Res 14, 1363-1369.

SC

Matsui, K., Tsume, Y., Amidon, G.E., Amidon, G.L., 2015. In Vitro Dissolution of Fluconazole and

NU

Dipyridamole in Gastrointestinal Simulator (GIS), Predicting in Vivo Dissolution and Drug-Drug Interaction Caused by Acid-Reducing Agents. Mol Pharm 12, 2418-2428.

MA

Matsui, K., Tsume, Y., Amidon, G.E., Amidon, G.L., 2016. The Evaluation of In Vitro Drug Dissolution of Commercially Available Oral Dosage Forms for Itraconazole in Gastrointestinal Simulator With

D

Biorelevant Media. J Pharm Sci 105, 2804-2814.

PT E

Mitra, A., Fadda, H.M., 2014. Effect of Surfactants, Gastric Emptying, and Dosage Form on Supersaturation of Dipyridamole in an in Vitro Model Simulating the Stomach and Duodenum. Mol Pharmaceut 11, 2835-2844.

CE

MOORE, #160, W., J., FLANNER, H., H., 1996. Mathematical comparison of dissolution profiles.

AC

Advanstar Communications, Duluth, MN, ETATS-UNIS. Moore, J.G., Christian, P.E., Brown, J.A., Brophy, C., Datz, F., Taylor, A., Alazraki, N., 1984. Influence of meal weight and caloric content on gastric emptying of meals in man. Dig Dis Sci 29, 513-519. Mudie, D.M., Murray, K., Hoad, C.L., Pritchard, S.E., Garnett, M.C., Amidon, G.L., Gowland, P.A., Spiller, R.C., Amidon, G.E., Marciani, L., 2014. Quantification of gastrointestinal liquid volumes and distribution following a 240 mL dose of water in the fasted state. Mol Pharm 11, 3039-3047. Murray, R., Eddy, D.E., Bartoli, W.P., Paul, G.L., 1994. Gastric emptying of water and isocaloric carbohydrate solutions consumed at rest. Med Sci Sports Exerc 26, 725-732. 39

ACCEPTED MANUSCRIPT Nicolaides, E., Symillides, M., Dressman, J.B., Reppas, C., 2001. Biorelevant dissolution testing to predict the plasma profile of lipophilic drugs after oral administration. Pharm Res 18, 380-388. Oberle, R.L., Chen, T.S., Lloyd, C., Barnett, J.L., Owyang, C., Meyer, J., Amidon, G.L., 1990. The influence of the interdigestive migrating myoelectric complex on the gastric emptying of liquids. Gastroenterology 99, 1275-1282.

PT

Psachoulias, D., Vertzoni, M., Goumas, K., Kalioras, V., Beato, S., Butler, J., Reppas, C., 2011.

two weak bases to fasted adults. Pharm Res 28, 3145-3158.

RI

Precipitation in and supersaturation of contents of the upper small intestine after administration of

SC

Ruff, A., Dressman, J.B., Kostewicz, E.S., 2013. Optimization of the transfer model to predict

NU

supersaturation and precipitation kinetics, AAPS - Poster presentation, San Antonio, USA. Schiller, C., Frohlich, C.P., Giessmann, T., Siegmund, W., Monnikes, H., Hosten, N., Weitschies, W.,

MA

2005. Intestinal fluid volumes and transit of dosage forms as assessed by magnetic resonance imaging. Aliment Pharmacol Ther 22, 971-979.

D

Shono, Y., Jantratid, E., Dressman, J.B., 2011. Precipitation in the small intestine may play a more

PT E

important role in the in vivo performance of poorly soluble weak bases in the fasted state: case example nelfinavir. Eur J Pharm Biopharm 79, 349-356. Steingoetter, A., Fox, M., Treier, R., Weishaupt, D., Marincek, B., Boesiger, P., Fried, M., Schwizer, W.,

CE

2006. Effects of posture on the physiology of gastric emptying: a magnetic resonance imaging study.

AC

Scand J Gastroenterol 41, 1155-1164. Sutton, S.C., 2009. Role of physiological intestinal water in oral absorption. AAPS J 11, 277-285. Takeuchi, S., Tsume, Y., Amidon, G.E., Amidon, G.L., 2014. Evaluation of a three compartment in vitro gastrointestinal simulator dissolution apparatus to predict in vivo dissolution. J Pharm Sci 103, 34163422. Taupitz, T., Dressman, J.B., Klein, S., 2013. In Vitro Tools for Evaluating Novel Dosage Forms of Poorly Soluble, Weakly Basic Drugs: Case Example Ketoconazole. J Pharm Sci-Us 102, 3645-3652.

40

ACCEPTED MANUSCRIPT Tonsberg, H., Holm, R., Bisgaard, J., Jacobsen, J., Mullertz, A., 2010. Effects of polysorbate 80 on the in-vitro precipitation and oral bioavailability of halofantrine from polyethylene glycol 400 formulations in rats. J Pharm Pharmacol 62, 63-70. Vertzoni, M., Pastelli, E., Psachoulias, D., Kalantzi, L., Reppas, C., 2007. Estimation of intragastric solubility of drugs: in what medium? Pharm Res 24, 909-917.

PT

Wagner, C., Jantratid, E., Kesisoglou, F., Vertzoni, M., Reppas, C., J, B.D., 2012. Predicting the oral absorption of a poorly soluble, poorly permeable weak base using biorelevant dissolution and

RI

transfer model tests coupled with a physiologically based pharmacokinetic model. Eur J Pharm

SC

Biopharm 82, 127-138.

NU

Wilding, I.R., Davis, S.S., Steed, K.P., Sparrow, R.A., Westrup, J., Hempenstall, J.M., 1994. Gastrointestinal transit of a drug-resinate administered as an oral suspension. International Journal

MA

of Pharmaceutics 101, 263-268.

Xu, S., Dai, W.G., 2013. Drug precipitation inhibitors in supersaturable formulations. Int J Pharm 453,

AC

CE

PT E

D

36-43.

41

ACCEPTED MANUSCRIPT

AC

CE

PT E

D

MA

NU

SC

RI

PT

Graphical Abstract

42