Contents lists available atScienceDirect
European Journal of Pharmaceutics and Biopharmaceutics
journal homepage:www.elsevier.com/locate/ejpbResearch paper
An in vitro dissolution–digestion–permeation assay for the study of
advanced drug delivery systems
Caroline Alvebratt
a,1, Janneke Keemink
a,1, Khadijah Edueng
a, Ocean Cheung
b, Maria Strømme
b,
Christel A.S. Bergström
a,c,⁎aDepartment of Pharmacy, Uppsala Biomedical Center, P.O. Box 580, Uppsala University, Uppsala SE-751 23, Sweden
bDivision of Nanotechnology and Functional Materials, Department of Materials Science and Engineering, Uppsala University, Uppsala SE-75121, Sweden cThe Swedish Drug Delivery Forum, Department of Pharmacy, Uppsala Biomedical Center, P.O. Box 580, Uppsala University, Uppsala SE-751 23, Sweden
A R T I C L E I N F O Keywords: Caco-2 cells Dissolution Digestion Permeation
Advanced drug delivery systems Lipid-based formulations
A B S T R A C T
Advanced drug delivery systems (ADDS) are widely explored to overcome poor aqueous solubility of orally administered drugs. However, the prediction of their in vivo performance is challenging, as in vitro models typically do not capture the interplay between processes occurring in the gut. In additions, different models are used to evaluate the different systems. We therefore present a method that allows monitoring of luminal pro-cessing (dissolution, digestion) and its interplay with permeation to better inform on the absorption of felodipine formulated as ADDS. Experiments were performed in a µFLUX-apparatus, consisting of two chambers, re-presenting the intestinal and serosal compartment, separated by Caco-2 monolayers. During dis-solution–digestion–permeation experiments, ADDS were added to the donor compartment containing simulated intestinal fluid and immobilized lipase. Dissolution and permeation in both compartments were monitored using in situ UV-probes or, when turbidity interfered the measurements, with HPLC analysis.
The method showed that all ADDS increased donor and receiver concentrations compared to the condition using crystalline felodipine. A poor correlation between the compartments indicated the need for an serosal compartment to evaluate drug absorption from ADDS. The method enables medium-throughput assessment of: (i) dynamic processes occurring in the small intestine, and (ii) drug concentrations in real-time.
1. Introduction
Oral drug administration is the most convenient route of adminis-tration, but is unfortunately problematic for many new drug candidates. Potent drug candidates often display unfavorable biopharmaceutical properties such as high lipophilicity and up to 70% are insufficiently soluble to allow complete absorption in the intestine[1]. To address this concern, a variety of advanced drug delivery systems (ADDS) have been developed. ADDS overcome solubility limitations by increasing the rate and extent of dissolution which thereby provides a greater concentration gradient to drive drug absorption from the intestinal lumen to the blood[2,3].
The compositions and manufacturing processes of ADDS vary sig-nificantly, and the most suitable choice of formulation strategy highly
depends on the properties of the drug. Frequently investigated ADDS include amorphous solid dispersions (ASDs), carrier-based delivery systems (CBDS), and lipid-based formulations (LBFs) [2]. ASDs and CBDS are typically used to stabilize the amorphous state of a drug, a state which is more soluble and dissolves faster than the crystalline state[4,5]. LBFs are used to deliver compounds predissolved in a liquid mixture of excipients. LBFs thereby increase intestinal drug con-centrations by circumventing the in vivo dissolution step, but also by keeping the drug solubilized in colloidal structures during its transfer through the gastrointestinal (GI) tract[6].
Literature studies report the potential and limitations of different ADDS[2,3,7]. Although this knowledge is useful, the prediction of the in vivo performance of these systems remains challenging[8]. ADDS performance depends on a complex interplay between physiological
https://doi.org/10.1016/j.ejpb.2020.01.010
Received 11 October 2019; Received in revised form 14 January 2020; Accepted 20 January 2020
Abbreviations: ADDS, Advanced drug delivery systems; ASD, Amorphous solid dispersion; CBDS, Carrier-based delivery systems; AUC, Area under the curve; FFA,
Free fatty acid; HBSS, Hanks’ Balanced Salt Solution; LBF, Lipid-based formulations; IIIB-LC, LBF type IIIB long chain; IIIB-MC, LBF type IIIB medium chain; IV, LBF type IV; MMC, Mesoporous magnesium carbonate; TPGS, d-α-Tocopheryl polyethylene glycol 1000 succinate; PVP, Polyvinylpyrrolidone
⁎Corresponding author.
E-mail address:christel.bergstrom@farmaci.uu.se(C.A.S. Bergström). 1These authors contributed equally to this work.
Available online 23 January 2020
0939-6411/ © 2020 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/BY-NC-ND/4.0/).
processes in the GI tract[9]. These processes are usually evaluated in various in vitro assays studying solubility, dissolution, digestion or permeation[10,11]. More elaborate in vitro models evaluate dissolu-tion and permeadissolu-tion simultaneously. These models comprise a donor and receiver compartment, to mimic the intestinal lumen and the ser-osal compartments. The two compartments are separated by a mem-brane representing the intestinal epithelium. Solubility and dissolution can be measured in the donor compartment and permeation is mon-itored on the receiver side[12–14].
In the current study, the µFLUX system was used. The µFLUX system is a small-scale dissolution-permeation setup that builds on the µDISS Profiler and allows close monitoring to time sensitive events, including supersaturation and precipitation, through in situ UV measurements of drug concentrations. The system has been proven useful in evaluating and ranking the performance of various enabling formulations (nano-suspensions and solid dispersions) containing itraconazole[13]. How-ever, for the assessment of drug release from LBFs, lipid digestion should be taken into consideration. Therefore, LBF performance is ty-pically evaluated in in vitro lipolysis studies to mimic the digestion in the intestine. In these studies, LBFs are dispersed in simulated intestinal fluids and digested with pancreatic enzymes. A pH-stat apparatus maintains the pH by titrating released free fatty acids to allow de-termination of the extent of digestion. Aqueous drug concentrations, assumed to be available for absorption, are used to predict in vivo ex-posure[6]. However, these predictions often result in a poor in vitro-in vivo correlation (IVIVC) [15–18]. Therefore, a semi-dynamic, small-scale method has recently been presented; the in vitro lipolysis was immediately followed by a drug permeation assay evaluating drug transfer across a biomimetic barrier [19]. Although, this method is useful for understanding the role of passive diffusion in drug permea-tion, methods that use more physiologically relevant membranes, in-line with the lipolysis assay would better capture the interplay between intraluminal processes. Such models include: (i) an in vitro lipolysis assay coupled to an in situ intestinal perfusion assay in rats; and (ii) an in vitro lipolysis-permeation model that allows the prediction of in vivo exposure of compound formulated with lipids. Despite their physiolo-gical relevance, both these experimental methods are time consuming and therefore more suitable for mechanistic studies than the method suggested herein[15,18,20].
The complexity of the intestinal environment and the significantly different release mechanisms of the various ADDS make it challenging to design an experimental setup that allows direct comparison of dif-ferent formulations strategies. This study aimed to develop an in vitro method to predict performance of a range of ADDS with regards to dissolution, digestion and permeation. The µFlux system was used in the µDiss Profiler to allow rapid and continuous in situ determinations of drug dissolution from various formulations and their permeation across a Caco-2 cell monolayer. The poorly water-soluble neutral drug felodipine (logP 4.8) was selected as a model compound[21]. 2. Materials and methods
2.1. Materials
All culture media and supplements were purchased from Invitrogen AB (Sweden). [14C]-mannitol was purchased from PerkinElmer Sverige AB (Sweden); Novozym® 435 (immobilized lipase) was obtained from Strem Chemicals (France); and fasted state simulated intestinal fluid (FaSSIF) powder was obtained from biorelevant.com (UK). Captex 355 and Capmul MCM, and felodipine were kind gifts from Abitec (USA) and AstraZeneca (Mölndal, Sweden), respectively. Lucifer yellow, acetonitrile and methanol were obtained from VWR (Sweden). 4-bro-mophenyl boronic acid, Cremophor EL, Soybean oil, Tween 85, Carbitol, sodium chloride, sodium hydroxide, Bis-Tris methane (Bis-Tris), Tris-maleate, calcium chloride dehydrate (CaCl2∙2H20), sodium acetate and dimethyl sulfoxide (DMSO) were purchased from Sigma
Aldrich (MO, USA). The nylon net filters (11 μm pore size), for use in the protective filters, were purchased from Merck Millipore (Billerica, MA, USA).
2.2. Formulations
2.2.1. Lipid-based formulations
Lipid-based formulations were prepared as described previously [22]. Briefly, excipients were pre-heated (37 °C) and weighed into glass vials according to predefined fractions (% w/w;Table 1). These vials were subsequently sealed, vortex-mixed and placed on a shaker, at 37 °C for 24 h. Felodipine (70 mg/g) was incorporated into the LBFs (Table 1). The required masses of model compound and LBFs were weighed into glass vials that were sealed, vortex-mixed and placed on a shaker, at 37 °C for an additional 24 h. Loading capacity was de-termined in triplicate as described before[22]. Drug loading in LBFs corresponded to 40, 32 and 30% of the total loading capacity for fe-lodipine in formulations IIIB-MC, IIIB-LC and IV, respectively. Felodi-pine concentrations in LBFs loaded in mesoporous magnesium carbo-nate (MMC) were twice as concentrated to obtain the same drug load in the final LBF-MMC formulations.
2.2.2. Mesoporous magnesium carbonate formulations
Mesoporous carriers e.g. mesoporous magnesium carbonate has been shown as a viable carrier for various drug molecules[23,24]. MMC with an average pore size of ~5 nm was synthesized according to the proce-dures described previously [25]. Felodipine was loaded in the meso-porous magnesium carbonate (MMC) formulation at a 1:3 w/w ratio using solvent evaporation, in accordance with a published protocol[5]. In short, 2 mg/mL crystalline felodipine was dissolved in ethanol, and then 6 mg/mL MMC was added. The suspension was stirred (500 rpm) at room temperature for 24 h, and then placed in an oven at 70 °C to evaporate the solvent during an additional 24 h. Modulated differential scanning ca-lorimetry confirmed that the felodipine loaded in MMC was amorphous (Fig. S1). Solid LBFs were obtained by simple mixing of felodipine-loaded
Table 1
Composition and drug load of formulations.
Formulation Composition (w/w) Drug load*
(mg/g) Amountformulation used**(g) IIIB-MC 50% Cremophor EL 70 0.375 25% Carbitol 12.5% Captex 355 12.5% Capmul MCM EP IIIB-LC 50% Carbitol 70 0.375 45% Tween85 5% Soybean oil IV 50% Cremophor EL 70 0.375 50% Carbitol IIIB-MC-MMC 50% IIIB-MC 70 0.375 50% MMC IIIB-LC-MMC 50% IIIB-LC 70 0.375 50% MMC IV-MMC 50% IV 70 0.375 50% MMC MMC 25% Crystalline drug 250 0.105 75% MMC
ASD 10% Crystalline drug 100 0.262 90% PVP
IIIB-MC; IIIB-LC; IV: Lipid based formulation type IIIB medium chain; type IIIB long chain; type IV. MMC: mesoporous magnesium carbonate. ASD: amorphous solid dispersion.
* mg felodipine in g formulation.
** g formulation used in 15 mL dissolution media the dissolution-digestion-permeation assay.
LBFs into the MMC (20 nm pore size) at a 1:1 w/w ratio. The LBF was added to the MMC, and stirred until all LBF had been adsorbed and a white free-flowing powder obtained.
2.2.3. Amorphous solid dispersion
The spray-drying solution was prepared by dissolving the crystalline felodipine and polyvinylpyrolidone K30 (PVP K30) in a mixture of ethanol:dichloromethane (50:50% w/w), giving a final ratio of 10:90 (% w/w drug:polymer)[26]. A Büchi Mini Spray Dryer B-290 (Swit-zerland) was used to produce the amorphous solid dispersion (ASD) with the following spray-drying conditions: inlet temperature (55 °C), aspiration rate (75%), and pump rate (4 mL/min)[27]. The ASD was further dried overnight to remove any residual solvent and then stored in a vacuum desiccator containing silica beads until analysis. Modu-lated differential scanning calorimetry confirmed that felodipine was amorphous in the ASD (Fig. S1).
2.3. Cell culture
Caco-2 cells were obtained from American Type Culture Collection (Manassas, Virginia) and cultivated as described previously [28]. Briefly, Caco-2 cells (passages 95 to 105) were suspended in Dulbecco’s modified Eagle’s medium supplemented with 10% fetal calf serum, 1% minimum essential medium nonessential amino acids, penicillin (100 U/mL), and streptomycin (100 µg/mL). Subsequently, cells were seeded on permeable polycarbonate filter supports (0.45-µm pore size; Trans-well Costar, Sigma-Aldrich) at a density of 440,000 and 320,000 cells/ cm2for 12- and 24- mm diameter inserts, respectively. Medium was changed every other day. Monolayers were used for experiments be-tween days 21 and 26 after seeding.
2.4. Transport studies
To evaluate the effect of the buffer on transepithelial transport, the permeabilities of felodipine and mannitol were determined across Caco-2 monolayers in: (i) Hanks’ Balanced Salt Solution (HBSS, pH 6.5) supplemented with 3.0 mM taurocholate and 0.75 mM lecithin, and (ii) modified lipolysis buffer (pH 6.5, containing 200 mM Bis-Tris, 4.5 mM CaCl2, 3.0 mM taurocholate and 0.75 mM lecithin). All solutions were pre-warmed to 37 °C. Prior to transport experiments, the cells were washed and then equilibrated (15 min) with HBSS (pH 7.4). Subsequently, the confluence of the cell monolayers was assessed by transepithelial electrical resistance measurements. Only monolayers with initial values greater than 250 Ω∙cm2 were used for transport studies. After the cells were equilibrated, the buffer was removed and the filters were transferred to wells containing 1.2 mL of HBSS (pH 7.4) with 0.2% of d-α-tocopheryl polyethylene glycol 1000 succinate (TPGS). The latter was used to maintain a sink in the receiver com-partment for enabling rapid translocation of felodipine across the ba-solateral membrane [29]. Transport studies were initiated by adding 400 µL to the apical chamber of: (i) HBSS supplemented with 3.0 mM taurocholate and 0.75 mM lecithin, or (ii) modified lipolysis buffer, both of which were spiked with felodipine (25 µM) and [14C] mannitol. Samples (600 µL) were removed from the basolateral chamber after 15, 30 and 60 min, and replaced with fresh HBSS (pH 7.4) containing TPGS. Samples were analyzed using a liquid scintillation counter (1900CA TriCarb; PerkinElmer Life Sciences) or HPLC-UV (seeSection 2.8).
The apparent permeability coefficient (Papp) was calculated ac-cording to the following equation:
= P dQ dtxAxC 1 app donor (1)
where Q is the amount of compound appearing in the receiver com-partment as a function of time (t); A is the surface area of the Transwell membrane (1.12 cm2); and C
donoris the initial compound concentration in the donor compartment.
2.5. In vitro lipolysis
In vitro lipolysis was carried out as described previously[30], with minor modifications. LBFs (1.25 g) or LBF-loaded MMC (2.5 g), were dispersed in either 50 mL lipolysis buffer (pH 6.5, containing 2 mM Tris–maleate, 1.4 mM CaCl2, 150 mM NaCl, 3.0 mM taurocholate and 0.75 mM lecithin), or modified lipolysis buffer, for 10 min in a ther-mostat-jacketed glass vessel (Metrohm, Switzerland) using a propeller stirrer (450 rpm). Digestion was initiated by addition of immobilized lipase (final concentration 125 PLU/mL). A pH-stat (Metrohm 907 Ti-trando) was used to: (i) maintain a pH of 6.5 through titration with NaOH for experiments performed in the lipolysis buffer, or (ii) monitor the pH during the experiments performed in the modified lipolysis buffer. The FFA release was determined based on the amount of NaOH used for titration (lipolysis buffer)[31], or with a kit for free fatty acid quantification (both buffer systems). The kit (MAK044, Sigma Aldrich, USA), was used according to the manufacturer’s instructions. In addi-tion, samples were withdrawn to evaluate the felodipine distribution across the different phases (i.e., aqueous, oil, and solid phases) in the vessel during digestion of the type IIIB-LC LBF. Samples were: (i) treated with 5 µL/mL lipase inhibitor (0.5 M 4-bromophenyl boronic acid in methanol) to inhibit further lipolysis, and (ii) vortexed and centrifuged (21,000g at 37 °C for 10 min) to separate the three phases. The phases were diluted 10-fold with acetonitrile, prior to further centrifugation (21,000g at 20 °C for 10 min), a second 10-fold dilution in mobile phase, and HPLC-UV analysis (seeSection 2.8).
2.6. Drug solubility in donor compartment
A small-scale shake flask method was used to determine the ap-parent solubility of felodipine in (undigested) donor medium [32]. Briefly, an excess of crystalline felodipine (~5 mg) was added to 1 mL of modified lipolysis buffer containing blank ADDS (i.e. no felodipine present) in the same concentration that was used during dis-solution–digestion–permeation experiments. After 24 h of shaking (300 rpm) at 37 °C, test tubes were centrifuged (2,300g at 37 °C, 10 min) and the supernatants were diluted in acetonitrile (1:1). Samples were analyzed by HPLC-UV (seeSection 2.8). It should be noted that solubility measurements in undigested media reflect a reference point as digestion could alter the solvation capacity of the system. Super-saturation for conditions containing digestible LBFs is therefore pos-sibly underestimated herein.
2.7. Dissolution–digestion–permeation measurements
All dissolution–permeation experiments were conducted in the µFLUX apparatus connected to the µDiss Profiler (Pion Inc., USA). The path lengths (2–20 mm) of the in situ UV-probes and the range of the standard curves were selected on the basis of the expected concentra-tions in the compartments (seeSection 2.6). Standard curves were es-tablished for each probe and each experimental condition with aliquots of a DMSO stock in: (i) modified lipolysis buffer containing relevant excipients (donor compartment), and (ii) HBSS containing 0.2% TPGS (pH 7.4, receiver compartment) to maintain sink conditions, as de-scribed previously[29,33]. Caco-2 monolayers (seeSection 2.3) were used as absorptive membranes (1.54 cm2surface area)[13]. Prior to the experiment, ADDS were weighed into the donor chamber (amounts equivalent to 1.75 mg/mL crystalline felodipine) and a protective nylon filter was placed on the in situ UV-probes to reduce scattering due to particle interference with the reading [5]. To initiate the dis-solution–digestion–permeation experiments (Fig. 1), 15 mL modified lipolysis buffer (pH 6.5)—spiked with 10 µM lucifer yellow used as integrity marker—was added to the donor chamber. For the receiver chamber, 15 mL of HBSS containing 0.2% TPGS (pH 7.4) was added. After 10 min, immobilized lipase (final concentration 125 PLU/mL) was added to the donor chamber. Although pancreatic extract is typically
used in digestion studies, it is incompatible with Caco-2 cells. Therefore immobilized lipase was used in the current study[30]. To minimize the interference during the in situ UV-measurements and increase the signal-to-noise ratio, analysis of the concentration profiles were per-formed using second derivatives, and a narrow interval of the wave-length was chosen (405–415 nm)[34,35]. The pH in both donor and receiver compartments was measured after each experiment.
During the dissolution–digestion–permeation experiment, samples of 100 µL were each taken from the donor and receiver compartments. Samples taken from the donor compartment were treated with lipase inhibitor (4-bromophenyl boronic acid) and immediately centrifuged (21,000g at 37 °C for 10 min) to inhibit further lipolysis and separate the oil, aqueous and solid phases. The aqueous phase was diluted 10-fold with acetonitrile and analyzed by HPLC-UV (seeSection 2.8). Lu-cifer yellow concentrations were measured in samples withdrawn from the receiver compartment and placed in a 96-well UV-plate reader (Tecan, Austria) to evaluate membrane integrity. Experiments were performed on two or more different days resulting in 4–6 data sets for each condition. Total exposure of felodipine in the donor and receiver compartments was evaluated using the area under the curve values (AUC) for the concentration vs. time profiles, which were calculated using the trapezoidal rule.
2.8. Sample preparation and HPLC analysis
HPLC-UV analysis was conducted with a Zorbraz Eclipse XDB-C18 column (4.6 × 100 mm) on a 1290 Infinity HPLC (Agilent Technologies). The injection volume was 20 μL. An isocratic mobile phase consisting of acetonitrile:sodium acetate buffer (25 mM, pH 5) at a ratio 80:20 (v/v) was used at a flow rate of 1 mL/min. UV absorbance was monitored at a wavelength of 360 nm. The retention time of fe-lodipine was 1.92 min.
2.9. Solid-state characterization 2.9.1. Polarized light microscopy
Drug formulations (Table 1) and donor media obtained after the dissolution–digestion–permeation experiments were characterized for their solid form using an Olympus BX51 microscope (Olympus, Japan) equipped with cross-polarizing filters. Samples were transferred to a microscope slide and images were recorded. Crystalline felodipine was examined as a reference.
2.9.2. Differential Scanning Calorimetry (DSC)
A DSC Q2000 Differential Scanning Calorimeter (TA Instrument Co., USA) was used to determine the thermal behavior of crystalline felo-dipine, felodipine-loaded MMC and the ASD (n = 3). Calibration of the DSC cell was performed using indium (melting temperature, Tm= 156.59 °C and heat of fusion, Hf= 28.57 J/g) and was purged with 50 mL/min of nitrogen. The melting point (Tm) of felodipine was determined by adding approximately 2.5 mg drug to an aluminum pan
that was sealed with an aluminum lid with pinholes. The samples were equilibrated at 0 °C, and heated at 10 °C /min to 180 °C. Modulated DSC (mDSC) was performed on the drug-loaded MMC and the ASD to con-firm the absence of crystalline drug. The formulations were weighed into the aluminum pans (5–6 mg) and each sample was equilibrated at 0 °C. It was then modulated at ± 0.5 °C every 60 s, and heated at 1 °C/ min to 200 °C. Onset values of Tmare reported for all measurements. 2.10. Statistical analysis
Statistical analysis was performed in GraphPad Prism 7 (GraphPad Software, USA). Student’s t-test was used to evaluate differences be-tween two groups. A one-way ANOVA, followed by a Tukey’s multiple comparison analysis test, was used to compare differences between more than two groups. P-values less than 0.05 were considered statis-tically significant. The supersaturation ratio was calculated for the donor compartments and the AUC for the supersaturation-time profiles was determined. The Spearman rank-order correlation coefficient was calculated to investigate the correlation between AUCs of the con-centration–time profiles obtained in the donor and receiver compart-ments and between the supersaturation-time curve in the donor com-partment and the concentration-time curve in the receiver compartment.
3. Results
3.1. Experimental conditions for the dissolution–digestion–permeation experiments
3.1.1. Modified lipolysis buffer and transepithelial transport
Transport experiments were performed with HBSS and modified lipolysis buffer to evaluate the effect of the modified lipolysis buffer on paracellular transport and passive diffusion. Paracellular transport of mannitol was unaffected by the type of buffer used. Moreover, apparent permeability (Papp) values were below 0.5 × 10-6cm/s, indicating that the Caco-2 monolayers remained confluent during the 60 min incuba-tion with this buffer (Fig. 2A). In addition, the model compound felo-dipine, which permeates through passive diffusion[36], did not show a statistically significant difference in Pappvalues following incubation in either HBSS or modified lipolysis buffer (Fig. 2B). We therefore con-cluded that Caco-2 cells were compatible with the modified lipolysis buffer.
3.1.2. Modified lipolysis buffer and lipid digestion
In the standard in vitro lipolysis assay[31], a pH value of 6.5 is maintained through constant titration of free fatty acids with NaOH. Since the µFLUX apparatus does not allow for continuous titration, a modified lipolysis buffer with high buffer capacity was used to maintain the pH value at a biorelevant level (i.e. 200 mM Bis-Tris in the modified lipolysis buffer vs 2 mM Tris Maleate in the standard buffer). Sup-porting Fig. S2 shows the pH-values during digestion of the LBFs and the LBF-loaded-MMC. A negligible drop in pH was observed during the digestion of the type IIIB-LC and IV formulations, which contain either a small amount (i.e. 5% w/w,Table 1) or no lipids, respectively. The digestion of the IIIB-MC formulation resulted in a pH drop to 6.4. During dispersion and digestion of the LBF-loaded-MMC, the buffer could not maintain the pH, which increased to 8.6 for all formulations (Fig. S2).
Type IIIB-MC formulation, the most digestible LBF in this study (Table 1), was digested in lipolysis buffer and in the modified lipolysis buffer. In addition, IIIB-MC-loaded MMC was digested in the modified lipolysis buffer. Digestion in the standard buffer was monitored through titration of released free fatty acids over time (Fig. 3A). Release of free fatty acids in all experiments (i.e. both the standard and modified li-polysis buffer) was determined with a quantification kit at the end of a 60-min digestion (Fig. 3B). The release values determined with the kit
were slightly higher than those obtained during titration (Fig. 3A and B, conditions shown in orange).Fig. 3B shows that neither modification of the buffer nor the loading of LBFs in MMC affects lipid digestion. 3.2. Dissolution–digestion–permeation
Felodipine concentrations measured in the µFLUX system during the dissolution–digestion–permeation experiments are depicted in Fig. 4. During experiments with the ASD and type IIIB-LC LBF, turbidity in the donor compartment prevented concentration measurements with the in situ UV-probes. Therefore, HPLC analysis was performed on samples collected from the donor compartment (Fig. 4., open symbols). Sup-porting Fig. S3 shows that both methods, i.e. analysis with the in situ UV-probes and HPLC, gave similar results.
Fig. S3 also shows the apparent solubility of felodipine in the modified lipolysis buffer containing relevant excipients, which allows the evaluation of supersaturation. The MMC and ASD show a transient supersaturation followed by precipitation. PLM results showed that precipitation occurred, at least partly, in the crystalline form (Fig. S4). The dispersion of the LC type IIIB LBF did not induce supersaturation since most of the felodipine resided in the oil phase upon dispersion and digestion (Fig. S5). The other LBFs induced supersaturation. This su-persaturation was maintained during the experiment with the MC type IIIB LBF (Figs. S3 and S4), yet was lost during the experiment with the type IV LBF, resulting in crystalline precipitate (Fig. S4). The LBF-loaded MMC formulations did not result in supersaturation during the experiments with type IIIB-LC- and IIIB-MC-loaded MMC. On the
contrary, type IV-loaded MMC induced a transient supersaturation followed by a relatively fast precipitation compared to the condition with only type IV LBF (Fig. S3).
Caco-2 monolayer integrity was evaluated using lucifer yellow. With the exception of the type IIIB-MC LBF, monolayer integrity was maintained during the entire experiment (i.e., 10 min of dispersion followed by 60 min of digestion). During experiments with type IIIB-MC LBF, monolayer integrity was lost after 30 min of digestion. To compare all ADDS, total felodipine exposure in the donor and receiver com-partments was calculated during the 10 min of dispersion and 30 min of digestion and using the donor concentrations obtained by HPLC ana-lysis (Fig. 5A and B). All ADDS showed solubility-enhancing effects that resulted in a relatively higher permeation than that produced by the crystalline felodipine (Figs. 4, 5A and 5B), however, for the MMC and IIIB-LC-MMC this was not statistically significant (Fig. 5B). The total permeation of felodipine formulated in the ASD resulted in a sig-nificantly higher AUC-value than values obtained with any other ADDS (p < 0.0001;Fig. 5B). A detailed list with significance levels com-paring the formulations shown inFig. 5A and B can be found in Table S1.Fig. 5C shows the rank order of the AUC-10-30obtained in the two compartments. A poor correlation was found between the concentration in the donor and receiver compartments, as indicated by a Spearman correlation of 0.61. No correlation was found between AUC’s of the supersaturation ratio-time curve in the donor compartment and the concentration-time curve in the receiver compartment (Spearman cor-relation = 0.42).
Fig. 2. Effect of modified lipolysis buffer on permeation across Caco-2 monolayers. A. Paracellular transport of mannitol, and B. Passive diffusion of felodipine. Bars
represent average apparent permeability (Papp) values ± SD (n = 3). No significant differences were observed after incubation of the compounds in Hanks’ balanced salt solution (HBSS) or modified lipolysis buffer, both used at pH 6.5.
Fig. 3. Free fatty acid (FFA) release during digestion of LBF type IIIB-MC. A. FFA titration during digestion of IIIB-MC in lipolysis buffer, and B. FFA release
determined with titration or the quantification kit after digestion of IIIB-MC in lipolysis buffer or modified lipolysis buffer and digestion of IIIB-MC-loaded MMC in modified lipolysis buffer. Values are presented as mean ± SD (n = 3). A lower FFA release was observed when titration was used, compared to the quantification kit (p-value < 0.01). No significant differences were observed between the conditions evaluated using the kit in panel 3B.
4. Discussion
The lack of in vitro methods that allows a head-to-head comparison of the intestinal performance of ADDS prior to in vivo studies is an important issue. The majority of compounds in the drug development pipeline show poor aqueous solubility and therefore require formula-tion strategies to enable absorpformula-tion. In this work, a small-scale method that simultaneously studies dissolution, digestion and permeation of felodipine across a Caco-2 cell monolayer in the µFLUX system from several ADDS was introduced, as ADDS performance highly depends on the interplay of all these processes.
For the in vitro method to be physiologically relevant, certain de-mands on the experimental method need to be considered. Caco-2 cells were chosen as an absorption membrane since differentiated Caco-2
cells have been shown to closely resemble the intestinal epithelium [28]. In addition, the presence of enzymes is required to digest LBF. However, such enzymes may be detrimental to the cells and previous lipolysis–permeation studies have therefore used immobilized enzyme with a Caco-2 monolayer as the absorptive membrane[18,30]. In vitro digestion with this non-specific lipase has been shown to result in (i) a complete digestion of mixed- or triglycerides[37]and (ii) a solvation capacity of digestion medium that is comparable to that observed in digestion medium aspirated from dog duodenums upon administration of blank LBFs not loaded with drug[20]. Caco-2 cells displayed lower tolerability to undigested MC formulations compared to LC and type IV-formulations in a static transwell model. Conversely, LC IV-formulations were more detrimental upon digestion[30]. This indicated that the type of LBF that is to be explored determines the length of the
Fig. 4. Felodipine concentrations in the donor and the receiver compartments of the µFLUX system during dissolution–digestion–permeation experiments. A and B
show the felodipine concentrations in the donor compartment and receiver compartment, respectively, during experiments with amorphous material formulated in an amorphous solid dispersion (ASD) or mesoporous magnesium carbonate (MMC). C and D show felodipine concentrations in the donor and receiver compartments, respectively, during experiments with the lipid-based formulations. E and F show felodipine concentrations in the donor and receiver compartments, respectively, during experiments with lipid-loaded-MMC. All panels show the results obtained with crystalline felodipine. Values are presented as mean ± SD (n ≥ 4). Closed symbols represent values obtained with in situ UV-probes and open symbols with HPLC analysis.
digestion–permeation experiment. In this work, the Caco-2 monolayer remained intact for all formulations during 60 min, with the exception of LBF type IIIB-MC for which integrity was lost after 30 min (Fig. S3). To address the pH changes related to the formation of free fatty acids during digestion, a Bis-Tris buffer was used that had a high buffer capacity (i.e. modified lipolysis buffer) [38]. As this buffer has not previously been used with the Caco-2 model, compatibility with the monolayers was evaluated by studying the permeability of mannitol and felodipine, dissolved in the modified lipolysis buffer or HBSS (Fig. 2). Our results are in agreement with those of a previous study by Alsenz and Haenel[39]. They used 40 mM Bis-Tris buffer (compared with the 200 mM used herein), but their modified lipolysis buffer was likewise compatible with Caco-2 monolayers and did not affect the permeation of the model compounds.
The pH-stabilizing capacity of our modified lipolysis buffer is con-sistent with what has been reported for LBFs containing similar lipid fractions[38]. The buffer maintained a pH of 6.5 for the types IIIB-LC and IV LBFs, containing 5 and 0% (w/w) lipids respectively, whereas there was a slight decrease during the digestion of the IIIB-MC (25% w/ w lipids) for which a higher extent of digestion occurs. Despite the high buffer capacity, there was a more pronounced shift in pH for the lipid-loaded MMC. This can be contributed to the partial dissolution of the MMC[5]. It is yet to be determined whether this pH effect occurs in vivo. Regardless, the impact of the pH on the absorption will be com-pound dependent. In our study, a neutral drug was used, so no sig-nificant change in solubility or permeation was expected.
The pH shift may significantly affect both the solubility and per-meability of ionizable compounds, depending on the physicochemical properties of the drug. In addition to the solubility and permeation of felodipine, the digestion capacity of the immobilized lipase was un-affected by the changes in pH (Fig. 3B). This is consistent with what has previously been reported by Phan et al.[37], who concluded that the enzymatic activity of the immobilized lipase is independent of pH and type of buffer used. The small discrepancy in the extent of digestion determined using titration and the free fatty acid kit could be attributed to the fact that the kit accounts for all free fatty acids, whereas only ionized free fatty acid is titrated with NaOH.
Apart from maintaining the pH, the Bis-Tris buffer has the ad-vantage that its intrinsic UV absorption is lower than for standard transport media used in permeability studies (e.g. HEPES and HBSS) [39]. This facilitates real-time measurement of felodipine concentra-tions using in situ UV-probes. However, buffering agents can affect the performance of the ADDS. The intestinal buffer comprises of bicarbo-nate buffer, making this the most in vivo relevant buffering system for studying intestinal performance in vitro[40]. However, the bicarbonate buffers are challenging to work with, because carbon dioxide needs to be purged continuously through the buffer to maintain the pH[41]. The
µFLUX system does not allow for this in the current setup. Thus, bi-carbonate may not be a suitable buffer system for this medium-throughput assay.
The ADDS selected for the study was commonly used in-house for-mulations, and was not optimized with regard to performance. The concentrations achieved in the donor compartments varied significantly across the ADDS, with the highest concentrations observed for the type IIIB-MC and type IV LBFs (Fig. 4). These high concentrations likely result from: (i) solubilization in the presence of colloidal structures, and (ii) increased solvation capacity in the presence of co-solvents and surfactants. These relatively high felodipine concentrations did not translate into high permeation. Instead, most formulations resulted in similar permeation (with the exception of the ASD). During experi-ments with type IIIB-LC LBF, aqueous drug concentrations were rather low due to partitioning of drug into the lipid droplets (Fig. S5).
A discrepancy between solubility and permeability has been ob-served on numerous occasions, because permeation is mainly depen-dent on the free aqueous drug concentrations[10,42,43]. A lipophilic drug typically partitions into the colloidal structures formed in the presence of solubilizing compounds (e.g., lipids and free fatty acids), reducing the total drug available for permeation[18]. Unfortunately, the UV-measurements do not discriminate between free and solubilized drug, which explains the weak correlation between the concentrations observed in the donor versus those in the receiver compartment (Fig. 5).
The ASD does not contain solubilizing components, and the highest absorption was observed for this formulation. The permeation of felo-dipine from the ASD displayed a two-phasic behavior, with a faster absorption rate during the first 20 min (from −10 to 10 min), which then decreased (Fig. 4B). This is in agreement with observations by Tsinman et al, who speculated that this is the result of initial super-saturation, followed by precipitation[13].Fig. 4A indeed shows a su-persaturation during the first 20 min of the experiment, after which the equilibrium solubility is reached. In solutions with a pre-dissolved polymer and felodipine, liquid-liquid phase separation has also been seen, creating a visibly turbid system. The liquid-liquid phase separa-tion may serve as a reservoir, maintaining the activity of the solusepara-tion and leading to higher absorption than a crystalline system[14].
Supersaturation was observed for most other ADDS (Fig. S3). The stability of the supersaturated state will significantly influence the permeation-enhancing effect[44]. Therefore, many ADDS contain ex-cipients that can kinetically stabilize supersaturation (i.e., polymers) [3]. In the current study, higher permeation was seen for the ASD containing a stabilizing agent (PVP), than during the experiment with drug-loaded MMC that contained no polymer (Table 1). In the absence of a polymer, the supersaturation induced by drug-loaded MMC was less stable than the polymer-containing ASD, and precipitation was
Fig. 5. Total exposure of felodipine in the donor and receiver compartments of the µFLUX system during dissolution–digestion–permeation experiments. A. Area
under the curve of concentration–time profiles during dispersion and 30 min of digestion in the donor compartment (AUC-10-30). B. Area under the curve of concentration–time profiles during dispersion and 30 min of digestion in the receiver compartment (AUC-10-30). C. Rank order of AUCs in the donor and receiver compartments; the formulation with the highest value is on top.
observed within minutes. In addition, the presence of solid MMC par-ticles may have induced nucleation, further driving the precipitation event. Drug supersaturation obtained with MMC has previously been stabilized with hydroxypropyl methylcellulose, suggesting that a polymer could have inhibited recrystallization and hence increased the permeation [45]. In addition to the effects on precipitation, the hy-drodynamics is an important factor to consider when studying per-meation in (digestion-) dissolution-perper-meation assays. The stirring in these systems should be optimized to minimize the aqueous boundary layer, i.e. the distance between the membrane and the location where the concentration becomes equal to the bulk concentration in the donor chamber[9].
Furthermore, the in situ UV-detection proved to be challenging, as the addition of excipients affected the UV-properties of the system. Major background scattering has been previously observed e.g., when spectroscopic techniques are used to study self-emulsifying drug de-livery systems [46]. This is therefore a limitation of the developed method. When model compounds or excipients are not compatible with real-time UV measurements, another method, e.g., HPLC-UV/MS, needs to be used to determine the concentration in the donor compartment. Many of the excipients selected in the current study displayed UV-ab-sorption at wavelengths < 300 nm. Since UV absorbance for felodi-pine can be monitored at high wavelengths (405–415 nm), the in situ monitoring could be utilized for most of the ADDS. Significant inter-ference with the UV-measurement was seen for LBFs containing long-chain lipids and the ASD. For both of these, the solution in the donor chamber was visibly milky, and for the ASD, powder had accumulated on the UV-probes by the end of the assay. In other words, the protective filter did not hinder the permeation of the small particles. Therefore, narrower pore sizes of the filter would be needed to prevent inter-ference.
The turbidity caused by the IIIB-LC LBF is likely due to the forma-tion of lipid droplets that reflect light and interfere with the UV-de-tection. Turbidity in LBF-dispersions has been seen for self-emulsifying drug delivery systems (SEDDS) with a droplet size greater than 0.25 µm [47]. These structures are smaller than the pore size of the nylon filter in this study and can therefore pass through this protective layer and reach the probe. For the IIIB-MC, a clear system formed upon dispersion of the LBF in the dissolution media, consistent with the formation of smaller droplet sizes (50–250 nm)[47,48].
The in situ UV-measurements allowed real-time monitoring of the drug concentrations in both the donor and the receiver compartments. We showed that these concentrations correlated well with the con-centrations determined by HPLC-UV (Fig. S3). However, an advantage with the in situ measurements is that concentrations can be measured often (every two seconds). This allows studies of processes that occur immediately upon exposure of the formulation to the dissolution medium or the digestive enzyme. Hundreds of data points can be col-lected within an hour, which enables more accurate monitoring of, e.g., supersaturation and precipitation, than when samples are withdrawn and analyzed separately. The advantage of the real-time measurement can be seen for several of the formulations for which high initial con-centrations and immediate precipitation occurred within minutes of the start of the assay (e.g., MMC in Fig. S3). By circumventing sampling and handling of samples, the physical integrity of the system is preserved which avoids redistribution of the drug across the different phases during e.g. centrifugation or filtration, which may produce inaccurate predictions of the aqueous drug concentrations in the system[46].
The aim of the current study was to develop a method that allows for simultaneous evaluation of dissolution, digestion and permeation for a range of ADDS varying in complexity and challenges in perfor-mance evaluation. It should be noted that the investigated formulations are commonly used in-house formulations that have not been optimized to increase permeation. To evaluate the applicability of the setup to predict in vivo exposure of drug molecules formulated in ADDS, an in vitro-in vivo correlation remains to be established.
5. Conclusion
In this work, we present an in vitro method that allows for real-time measurement of drug concentrations in the mimicked intestinal and blood compartments in response to processes in the gastrointestinal tract (dissolution, digestion, permeation). Simulated intestinal fluid with a strong buffer capacity and protective filters were used in the donor compartment of the µFLUX system to evaluate a range of ADDS. When excipients influenced the UV-absorptive properties of the dis-persed formulation, additional HPLC-UV analysis of samples from the donor compartment was used to measure aqueous felodipine con-centrations. A weak correlation was found between concentrations in the donor compartment and permeation to the receiver compartment, emphasizing the importance of in vitro methods that evaluates per-meation simultaneously with dissolution and digestion as suggested in this study. The method presented here is relatively rapid and simple, and could provide useful predictions during ADDS development, prior to in vivo studies.
Acknowledgements
This work received support from the European Research Council [Grant 638965], the Swedish Research Council [Grant 2014-3929], the Erling Persson Family Foundation and is part of an associated research project of the Swedish Drug Delivery Forum (SDDF).
Appendix A. Supplementary material
Supplementary data to this article can be found online athttps:// doi.org/10.1016/j.ejpb.2020.01.010.
References
[1] F. Carrière, Impact of gastrointestinal lipolysis on oral lipid-based formulations and bioavailability of lipophilic drugs, Biochimie 125 (2016) 297–305,https://doi.org/ 10.1016/j.biochi.2015.11.016.
[2] S.Y.K. Fong, A. Bauer-Brandl, M. Brandl, Oral bioavailability enhancement through supersaturation: an update and meta-analysis, Expert Opin. Drug Deliv. 14 (2017) 403–426,https://doi.org/10.1080/17425247.2016.1218465.
[3] J. Brouwers, M.E. Brewster, P. Augustijns, Supersaturating drug delivery systems: the answer to solubility-limited oral bioavailability? J. Pharm. Sci. 98 (2009) 2549–2572,https://doi.org/10.1002/jps.21650.
[4] G. Van den Mooter, The use of amorphous solid dispersions: A formulation strategy to overcome poor solubility and dissolution rate, Drug Discov. Today Technol. 9 (2012) e79–e85,https://doi.org/10.1016/j.ddtec.2011.10.002.
[5] C. Alvebratt, O. Cheung, M. Strømme, C.A.S. Bergström, A modified in situ method to determine release from a complex drug carrier in particle-rich suspensions, AAPS PharmSciTech. 19 (2018) 2859–2865, https://doi.org/10.1208/s12249-018-1024-1.
[6] O.M. Feeney, M.F. Crum, C.L. McEvoy, N.L. Trevaskis, H.D. Williams, C.W. Pouton, W.N. Charman, C.A.S. Bergström, C.J.H. Porter, 50years of oral lipid-based for-mulations: Provenance, progress and future perspectives, Adv. Drug. Deliv. Rev. 101 (2016) 167–194,https://doi.org/10.1016/j.addr.2016.04.007.
[7] M. Yang, W. Gong, Y. Wang, L. Shan, Y. Li, C. Gao, Bioavailability improvement strategies for poorly water-soluble drugs based on the supersaturation mechanism: an update, J. Pharm. Pharm. Sci. 19 (2016) 208–225https://doi.org/10.18433/ J3W904.
[8] E.S. Kostewicz, B. Abrahamsson, M. Brewster, J. Brouwers, J. Butler, S. Carlert, P.A. Dickinson, J. Dressman, R. Holm, S. Klein, J. Mann, M. McAllister, M. Minekus, U. Muenster, A. Müllertz, M. Verwei, M. Vertzoni, W. Weitschies, P. Augustijns, In vitro models for the prediction of in vivo performance of oral dosage forms, Eur. J. Pharm. Sci. 57 (2014) 342–366,https://doi.org/10.1016/j.ejps.2013.08.024. [9] A. Dahan, J.M. Miller, The solubility-permeability interplay and its implications in
formulation design and development for poorly soluble drugs, AAPS J. 14 (2012) 244–251,https://doi.org/10.1208/s12248-012-9337-6.
[10] A. Beig, J.M. Miller, D. Lindley, R.A. Carr, P. Zocharski, R. Agbaria, A. Dahan, Head-to-head comparison of different solubility-enabling formulations of etoposide and their consequent solubility-permeability interplay, J. Pharm. Sci. 104 (2015) 2941–2947,https://doi.org/10.1002/jps.24496.
[11] A. Dahan, A. Hoffman, The effect of different lipid based formulations on the oral absorption of lipophilic drugs: The ability of in vitro lipolysis and consecutive ex vivo intestinal permeability data to predict in vivo bioavailability in rats, Eur. J. Pharm. Biopharm. 67 (2007) 96–105,https://doi.org/10.1016/j.ejpb.2007.01.017. [12] M. Kataoka, K. Yano, Y. Hamatsu, Y. Masaoka, S. Sakuma, S. Yamashita, Assessment
permeation system, Eur. J. Pharm. Biopharm. 85 (2013) 1317–1324,https://doi. org/10.1016/j.ejpb.2013.06.018.
[13] K. Tsinman, O. Tsinman, R. Lingamaneni, S. Zhu, B. Riebesehl, A. Grandeury, M. Juhnke, B. Van Eerdenbrugh, Ranking itraconazole formulations based on the flux through artificial lipophilic membrane, Pharm. Res. 35 (2018) 161,https://doi. org/10.1007/s11095-018-2440-3.
[14] S.A. Raina, G.G.Z. Zhang, D.E. Alonzo, J. Wu, D. Zhu, N.D. Catron, Y. Gao, L.S. Taylor, Enhancements and limits in drug membrane transport using super-saturated solutions of poorly water soluble drugs, J. Pharm. Sci. 103 (2014) 2736–2748,https://doi.org/10.1002/jps.23826.
[15] M.F. Crum, N.L. Trevaskis, H.D. Williams, C.W. Pouton, C.J.H. Porter, A new in vitro lipid digestion – in vivo absorption model to evaluate the mechanisms of drug absorption from lipid-based formulations, Pharm. Res. 33 (2016) 970–982,https:// doi.org/10.1007/s11095-015-1843-7.
[16] N. Thomas, K. Richter, T.B. Pedersen, R. Holm, A. Müllertz, T. Rades, In vitro li-polysis data does not adequately predict the in vivo performance of lipid-based drug delivery systems containing fenofibrate, AAPS J. 16 (2014) 539–549,https://doi. org/10.1208/s12248-014-9589-4.
[17] B.T. Griffin, M. Kuentz, M. Vertzoni, E.S. Kostewicz, Y. Fei, W. Faisal, C. Stillhart, C.M. O’Driscoll, C. Reppas, J.B. Dressman, Comparison of in vitro tests at various levels of complexity for the prediction of in vivo performance of lipid-based for-mulations: Case studies with fenofibrate, Eur. J. Pharm. Biopharm. 86 (2014) 427–437,https://doi.org/10.1016/j.ejpb.2013.10.016.
[18] J. Keemink, E. Mårtensson, C.A.S. Bergström, Lipolysis-permeation setup for si-multaneous study of digestion and absorption in vitro, Mol. Pharm. 16 (2019) 921–930,https://doi.org/10.1021/acs.molpharmaceut.8b00811.
[19] H.A. Bibi, R. Holm, A. Bauer-Brandl, Simultaneous lipolysis/permeation in vitro model, for the estimation of bioavailability of lipid based drug delivery systems, Eur. J. Pharm. Biopharm. 117 (2017) 300–307,https://doi.org/10.1016/j.ejpb. 2017.05.001.
[20] L.C. Alskär, A. Parrow, J. Keemink, P. Johansson, B. Abrahamsson,
C.A.S. Bergström, Effect of lipids on absorption of carvedilol in dogs: Is coadmi-nistration of lipids as efficient as a lipid-based formulation? J. Control. Release. 304 (2019) 90–100,https://doi.org/10.1016/j.jconrel.2019.04.038.
[21] J.H. Fagerberg, C.A.S. Bergström, Intestinal solubility and absorption of poorly water soluble compounds: predictions, challenges and solutions, Ther. Deliv. 6 (2015) 935–1359,https://doi.org/10.4155/tde.15.45.
[22] L.C. Alskär, C.J.H. Porter, C.A.S. Bergström, Tools for early prediction of drug loading in lipid-based formulations, Mol. Pharm. 13 (2016) 251–261,https://doi. org/10.1021/acs.molpharmaceut.5b00704.
[23] P. Zhang, J. Forsgren, M. Strømme, Stabilisation of amorphous ibuprofen in Upsalite, a mesoporous magnesium carbonate, as an approach to increasing the aqueous solubility of poorly soluble drugs, Int. J. Pharm. 472 (2014) 185–191,
https://doi.org/10.1016/j.ijpharm.2014.06.025.
[24] Peng Zhang, Teresa Zardán Gómez de la Torre, Ken Welch, Christel Bergström, Maria Strømme, Supersaturation of poorly soluble drugs induced by mesoporous magnesium carbonate, Eur. J. Pharmaceut. Sci. 93 (2016) 468–474,https://doi. org/10.1016/j.ejps.2016.08.059.
[25] O. Cheung, P. Zhang, S. Frykstrand, H. Zheng, T. Yang, M. Sommariva, X. Zou, M. Strømme, Nanostructure and pore size control of template-free synthesised mesoporous magnesium carbonate, RSC Adv. 6 (2016) 74241–74249,https://doi. org/10.1039/C6RA14171D.
[26] D.E. Alonzo, Y. Gao, D. Zhou, H. Mo, G.G.Z. Zhang, L.S. Taylor, Dissolution and precipitation behavior of amorphous solid dispersions, J. Pharm. Sci. 100 (2011) 3316–3331,https://doi.org/10.1002/jps.22579.
[27] K. Edueng, D. Mahlin, P. Larsson, C.A.S. Bergström, Mechanism-based selection of stabilization strategy for amorphous formulations: Insights into crystallization pathways, J. Control. Release. 256 (2017) 193–202,https://doi.org/10.1016/j. jconrel.2017.04.015.
[28] I. Hubatsch, E.G.E. Ragnarsson, P. Artursson, Determination of drug permeability and prediction of drug absorption in Caco-2 monolayers, Nat. Protoc. 2 (2007) 2111–2119,https://doi.org/10.1038/nprot.2007.303.
[29] P. Berben, A. Bauer-Brandl, M. Brandl, B. Faller, G.E. Flaten, A.-C. Jacobsen, J. Brouwers, P. Augustijns, Drug permeability profiling using cell-free permeation tools: Overview and applications, Eur. J. Pharmaceut. Sci. 119 (2018) 219–233,
https://doi.org/10.1016/j.ejps.2018.04.016.
[30] J. Keemink, C.A.S. Bergström, Caco-2 cell conditions enabling studies of drug ab-sorption from digestible lipid-based formulations, Pharm. Res. 35 (2018) 74,
https://doi.org/10.1007/s11095-017-2327-8.
[31] Hywel D. Williams, Philip Sassene, Karen Kleberg, Jean-Claude Bakala-N'Goma,
Marilyn Calderone, Vincent Jannin, Annabel Igonin, Anette Partheil, Delphine Marchaud, Eduardo Jule, Jan Vertommen, Mario Maio, Ross Blundell, Hassan Benameur, Frédéric Carrière, Anette Müllertz, Christopher J.H. Porter, Colin W. Pouton, Toward the establishment of standardized in vitro tests for lipid-based formulations, Part 1: method parameterization and comparison of in vitro digestion profiles across a range of representative formulations, J. Pharmaceut. Sci. 101 (9) (2012) 3360–3380,https://doi.org/10.1002/jps.23205.
[32] C.A.S. Bergström, U. Norinder, K. Luthman, P. Artursson, Experimental and com-putational screening models for prediction of aqueous drug solubility, Pharm. Res. 19 (2002) 182–188,https://doi.org/10.1023/A:1014224900524.
[33] S.B.E. Andersson, C. Alvebratt, C.A.S. Bergström, Controlled suspensions enable rapid determinations of intrinsic dissolution rate and apparent solubility of poorly water-soluble compounds, Pharm. Res. 34 (2017) 1805–1816,https://doi.org/10. 1007/s11095-017-2188-1.
[34] S.B.E. Andersson, C. Alvebratt, J. Bevernage, D. Bonneau, C. da Costa Mathews, R. Dattani, K. Edueng, Y. He, R. Holm, C. Madsen, T. Müller, U. Muenster, A. Müllertz, K. Ojala, T. Rades, P. Sieger, C.A.S. Bergström, Interlaboratory vali-dation of small-scale solubility and dissolution measurements of poorly water-so-luble drugs, J. Pharm. Sci. 105 (2016) 2864–2872,https://doi.org/10.1016/j.xphs. 2016.03.010.
[35] K. Tsinman, A. Avdeef, O. Tsinman, D. Voloboy, Powder dissolution method for estimating rotating disk intrinsic dissolution rates of low solubility drugs, Pharm Res. 26 (2009) 2093–2100,https://doi.org/10.1007/s11095-009-9921-3. [36] T.L. Farrell, L. Poquet, T.P. Dew, S. Barber, G. Williamson, Predicting phenolic acid
absorption in caco-2 cells: a theoretical permeability model and mechanistic study, Drug Metab. Dispos. 40 (2012) 397–406,https://doi.org/10.1124/dmd.111. 041665.
[37] S. Phan, S. Salentinig, A. Hawley, B.J. Boyd, Immobilised lipase for in vitro lipolysis experiments, J. Pharm. Sci. 104 (2015) 1311–1318,https://doi.org/10.1002/jps. 24327.
[38] M.D. Mosgaard, P. Sassene, H. Mu, T. Rades, A. Müllertz, Development of a high-throughput in vitro intestinal lipolysis model for rapid screening of lipid-based drug delivery systems, Eur. J. Pharm. Biopharm. 94 (2015) 493–500,https://doi.org/10. 1016/j.ejpb.2015.06.028.
[39] J. Alsenz, E. Haenel, Development of a 7-day, 96-well Caco-2 permeability assay with high-throughput direct UV compound analysis, Pharm. Res. 20 (2003) 1961–1969.
[40] F.J.O. Varum, H.A. Merchant, A. Goyanes, P. Assi, V. Zboranová, A.W. Basit, Accelerating the dissolution of enteric coatings in the upper small intestine: Evolution of a novel pH 5.6 bicarbonate buffer system to assess drug release, Int. J. Pharm. 468 (2014) 172–177,https://doi.org/10.1016/j.ijpharm.2014.04.019. [41] B.J. Krieg, S.M. Taghavi, G.L. Amidon, G.E. Amidon, In vivo predictive dissolution:
comparing the effect of bicarbonate and phosphate buffer on the dissolution of weak acids and weak bases, J. Pharm. Sci. 104 (2015) 2894–2904,https://doi.org/ 10.1002/jps.24460.
[42] D. Porat, A. Dahan, Active intestinal drug absorption and the solubility-perme-ability interplay, Int. J. Pharm. 537 (2018) 84–93,https://doi.org/10.1016/j. ijpharm.2017.10.058.
[43] P. Berben, J. Brouwers, P. Augustijns, The artificial membrane insert system as predictive tool for formulation performance evaluation, Int. J. Pharm. 537 (2018) 22–29,https://doi.org/10.1016/j.ijpharm.2017.12.025.
[44] J. Bevernage, J. Brouwers, M.E. Brewster, P. Augustijns, Evaluation of gastro-intestinal drug supersaturation and precipitation: Strategies and issues, Int. J. Pharm. 453 (2013) 25–35,https://doi.org/10.1016/j.ijpharm.2012.11.026. [45] Jiaojiao Yang, Caroline Alvebratt, Peng Zhang, Teresa Zardán Gómez de la Torre,
Maria Strømme, Christel A.S. Bergström, Ken Welch, Enhanced release of poorly water-soluble drugs from synergy between mesoporous magnesium carbonate and polymers, Int. J. Pharmaceut. 525 (1) (2017) 183–190,https://doi.org/10.1016/j. ijpharm.2017.04.018.
[46] A. Bernkop-Schnürch, A. Jalil, Do drug release studies from SEDDS make any sense? J. Control. Release. 271 (2018) 55–59,https://doi.org/10.1016/j.jconrel.2017.12. 027.
[47] C.W. Pouton, Formulation of poorly water-soluble drugs for oral administration: physicochemical and physiological issues and the lipid formulation classification system, Eur. J. Pharm. Sci. 29 (2006) 278–287,https://doi.org/10.1016/j.ejps. 2006.04.016.
[48] C.W. Pouton, Lipid formulations for oral administration of drugs: non-emulsifying, self-emulsifying and ‘self-microemulsifying’ drug delivery systems, Eur. J. Pharm. Sci. 11 (2000) S93–S98,https://doi.org/10.1016/S0928-0987(00)00167-6.
