Dissolution behavior of various drugs in different FaSSIF versions

Journal Pre-proof

Dissolution behavior of various drugs in different FaSSIF versions Lukas Klumpp , Mathew Leigh , Jennifer Dressman PII: DOI: Reference:

S0928-0987(19)30411-7 https://doi.org/10.1016/j.ejps.2019.105138 PHASCI 105138

To appear in:

European Journal of Pharmaceutical Sciences

Received date: Revised date: Accepted date:

7 August 2019 24 October 2019 3 November 2019

Please cite this article as: Lukas Klumpp , Mathew Leigh , Jennifer Dressman , Dissolution behavior of various drugs in different FaSSIF versions, European Journal of Pharmaceutical Sciences (2019), doi: https://doi.org/10.1016/j.ejps.2019.105138

This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. 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. © 2019 Published by Elsevier B.V.

Dissolution behavior of various drugs in different FaSSIF versions Lukas Klumpp1, Mathew Leigh2, Jennifer Dressman1+ 1 2

Institute of Pharmaceutical Technology, Goethe University, Frankfurt am Main, Germany Biorelevant.com, London, United Kingdom

Keywords: Biorelevant Media, Dissolution, Human Intestinal Media, FaSSIF, Solubility, BCS II

+

Corresponding author:

Jennifer Dressman Institute of Pharmaceutical Technology Goethe University, Max-von-Laue-Str. 9, 60438 Frankfurt am Main, Germany. E-mail: [email protected]

Abstract Biorelevant media have proven to be useful in predicting the performance of poorly soluble drugs in the gastrointestinal tract. Several versions of fasted state simulated intestinal fluids have been published and compared with respect to their physical chemical properties and solubilization of drugs. However, to date there have been no reports in the literature comparing dissolution of poorly soluble drugs in these media. In this study eleven BCS Class II compounds (five nonionized compounds, three weak bases and three weak acids) were investigated with respect to their thermodynamic solubility and dissolution behavior in three biorelevant media simulating conditions in the small intestine (FaSSIF V1, FaSSIF V2 and FaSSIF V3). It was shown that the maximum percentage release of drugs from their commercial formulations can differ from the results for the thermodynamic solubility of the pure drug; these differences can be largely attributed to API presentation, composition of the formulation and manufacturing effects. 1

The results were additionally compared with data for solubility in HIF taken from the literature in order to determine which version of FaSSIF most closely corresponds to the physiological conditions. The different versions of FaSSIF are able to achieve solubility results similar to those in HIF, with closest results generally achieved in FaSSIF V1. The magnitude of solubility/dissolution differences among the three FaSSIF versions is dependent on the drug’s characteristics. In the case of weakly basic compounds, the differences among the FaSSIF versions are minor. For weakly acidic compounds the behavior in the different versions is primarily pH dependent and influenced by the lipid composition of the FaSSIF only to a minor extent. The differences in solubility and dissolution of the nonionized compounds among the three versions of FaSSIF becomes apparent above a log P value of 2.5, with larger differences among the versions at high log P values.

1. Introduction For more than two decades biorelevant media have been used to simulate the human gastrointestinal tract (Dressman et al., 1998; Galia et al., 1998). Due to their more physiologically relevant composition, they are superior for this purpose to compendial media, such as SGF or SIF (Otsuka et al., 2013), which are frequently used in dissolution testing for quality control purposes. The choice of biorelevant media for a given experiment depends on both the drug and formulation type as well as the conditions under which the dosage form is administered. Markopoulos and Andreas et al. classified dissolution media into various “Levels” of biorelevance, which makes it possible to select the most appropriate medium for a given active pharmaceutical ingredient (API) and formulation (Markopoulos et al., 2015). Especially when testing drug products containing poorly soluble drugs, it is important to choose a medium that not only simulates the pH and the buffer capacity of the gastrointestinal (GI) tract, but also contains bile components (e.g. bile salts, phospholipids, cholesterol) and (when appropriate) fat digestion products, which combine to reduce surface tension and form micellar structures. A

2

particularly important role is played by the small intestine media, since efficient absorption of the API into the body in most cases depends on how much of the API is in solution in this segment of the GI tract. Since the first development of Fasted State Simulated Intestinal Media (FaSSIF V1) in 1998 by Dressman et al. there have been two further versions, FaSSIF V2 and FaSSIF V3, which each differ in composition from the previous version (Fuchs et al., 2015; Galia et al., 1998; Jantratid et al., 2008). Especially in the third version (FaSSIF V3), the complexity of the composition is greater than in FaSSIF V1 and V2 and indeed it has been shown that the versions differ in terms of physicochemical properties such as surface tension, pH and osmolality (Klumpp et al., 2019). In addition, although the colloidal structures in the various media all correspond to those in human intestinal fluid (HIF), their relative prevalence differs. Thus the question arises as to which of the three versions most reliably produces results closely reflecting the behavior of the API and its dosage form in vivo. Although Fuchs et al. compared the solubility of a range of APIs in the three FaSSIF versions (Fuchs, 2015), no data directly comparing dissolution results in all three media have appeared in the literature to date. To investigate this question further, the solubility and dissolution behavior of eleven BCS Class II drugs were compared in the three FaSSIF versions. Where possible, the solubilities of the respective active substance in HIF and the FaSSIF versions were compared and in all cases the release behavior among the FaSSIF versions was examined. 2. Materials and Method 2.1. Chemicals and reagents All chemicals used were of analytical grade. Pure substance standards for the APIs were of secondary standard quality or higher. The commercial pharmaceutical products were ordered from Phoenix Pharma SE (Mannheim, Germany). The instant powders FaSSIF V1, FaSSIF V2 and FaSSIF V3 were kindly donated by Biorelevant.com Ltd. (London, United Kingdom). FaSSGF, FaSSIF V1 and FaSSIF V2 were prepared using the standard operating procedures provided by Biorelevant.com Ltd. (Biorelevant.com). FaSSIF V3 was prepared as described by Fuchs et al. (Fuchs, 2015). To afford a 3

better comparison of FaSSIF V3 with FaSSIF V1 and V2 (especially for ionizable APIs), the pH was lowered from 6.7 to 6.5 in certain experiments, but no other modifications to the composition were made. After preparation, the media were held for 2h at room temperature for equilibration. Compositions of the media are shown in Table 1.

Table 1. Compositions of biorelevant media used for solubility and dissolution studies

2.2. Active Pharmaceutical Ingredients Eleven APIs were investigated. Of these, five are nonionized compounds, three are weak bases and three are weak acids. They are all marketed as immediate release tablets or immediate release capsules. A list of the different APIs according to ionization characteristics, formulation and log P value can be found in Table 2.

Table 2. APIs investigated, with details of the commercial product tested, manufacturer, formulation type, classification by ionization and log P value. *Phenytoin is not ionized over the pH range studied (pH 6.5 – 6.7)

2.3. Solubility measurements Solubility tests were performed for ciprofloxacin, dipyridamole, ibuprofen, indinavir, phenytoin and zafirlukast in FaSSIF V1, V2 and V3. These were performed in UniPrep® filter devices (Whatman, Inc., USA), as a miniaturized version of the shake-flask method (Glomme et al., 2005). Briefly, ~20mg active substance was added to 3.0mL of the respective medium in a UniPrep® device and maintained at 24h at 37°C on an orbital shaker (Heidolph Instruments, Germany) (n = 3). After filtration through the integrated 0.45µm PTFE filter in the Uniprep®, the sample was immediately diluted with mobile phase to prevent precipitation and then analyzed by HPLC (see Table 3 for HPLC details). 2.4. Dissolution testing 4

Dissolution tests were performed in USP apparatus II (paddle, DT 720 from Erweka GmbH, Heusenstamm, Germany) with n = 6 vessels and 500mL medium per vessel. The paddle speed was set to 75 rpm and the temperature of the medium was maintained at 37 ± 0.5°C. 5mL samples were withdrawn manually after 5, 10, 15, 20, 30, 45, 60, 90 and 120 minutes. The sample was filtered through a 0.45µm syringe PTFE filter (WhatmanTM, Dassel, Germany) and the first 4mL were returned to the respective dissolution vessel, while the remaining 1mL was transferred into an Eppendorf tube and immediately diluted with mobile phase to prevent precipitation prior to HPLC analysis. Dissolution tests were performed in FaSSIF V1, FaSSIF V2 and FaSSIF V3 for each drug product. Further, weakly acidic APIs were tested in FaSSIF V3 at a reduced pH of 6.5 to evaluate the influence of pH on dissolution rate. If coning occurred, the test was repeated in a peak vessel. To determine the influence of the formulation on dissolution performance, dissolution tests with the pure API were performed for three compounds (carbamazepine, celecoxib and fenofibrate) in FaSSIF V1. The dissolution tests were performed in the same way as for the drug products, except that samples were taken after 5, 15, 30, 60 and 120 minutes. The other compounds were not investigated since either the salt forms differ between pure API and the drug product (ciprofloxacin and indinavir), the API in the drug product was amorphous (zafirlukast), the dose of the API was insufficient to reach thermodynamic solubility (ibuprofen) or the differences between thermodynamic solubility and maximum concentration achieved in the dissolution test were trivial. 2.5. HPLC analysis All HPLC methods were based on literature data (Berben et al., 2019; Dhabu and Akamanchi, 2002; Fuchs, 2015; Lobenberg et al., 2000; Mann et al., 2017; Mitra and Fadda, 2014; Rajadhyaksha et al., 2007; Shah and Ogger, 1986; Wu et al., 2008) and adjusted as necessary to obtain a short run time while maintaining a well-defined peak. The settings for the individual HPLC methods are presented in Table 3.

5

Table 3. HPLC setup for the analysis of each compound

2.6. Statistical analysis The statistical analysis of the dissolution results was performed using MS Excel 2016 (Redmond, United States) and SPSS Statistics Version 24 from IBM (Armonk, United States). The profiles were divided, where appropriate, into a region which could be described with first order kinetics and a plateau region (Fig. 1). If no plateau was reached in the profile, first order kinetics were used to model the entire dissolution profile. Each set of six kinetic constants (one per vessel) was checked for normality of the distribution of values and tested for significance using one-way ANOVA, with the Type I error set to 0.05. P-values are only reported where a significant difference (p ≤ 0.05) was reached, in which case a post-hoc Bonferroni test was additionally performed.

Figure 1. Fitting of the dissolution results. The profile which was divided into 2 regions for statistical analysis, a first order region (short lines) and, where appropriate, a plateau region (straight line). The example shown is for Celecoxib tablets.

3. Results 3.1. Solubility For five APIs the solubility in the FaSSIF versions had been measured by Fuchs et al. (Fuchs, 2015). For the remaining five compounds, the solubility was determined experimentally and the results are presented in Table 4, together with the data from Fuchs et al..

Table 4. Solubility (µg/mL) of 11 APIs in FaSSIF V1, V2 and V3. Solubility of APIs* had been measured by Fuchs et al.(Fuchs, 2015).

3.2. Dissolution 3.2.1.Nonionized compounds

6

Carbamazepine: Carbamazepine (Carbamazepine 200mg – 1A Pharma) showed similar dissolution in FaSSIF V1, FaSSIF V2 and FaSSIF V3. After fitting the profiles with a first-order model, one-way ANOVA indicated no significant differences in dissolution among the three media. After 120 minutes of dissolution testing, between 50% and 60% of the carbamazepine, corresponding to concentrations of 200 - 240 µg/mL, had been released from the tablets. The pure API dissolved to an extent of just over 71% in the same period, corresponding to a concentration of 287.7 ± 10.5µg/mL (Fig. 2a). However, concentrations did not reach the thermodynamic solubility in any of the media (295.3µg/mL - 311.7 µg/mL, see Table 4). Both the thermodynamic solubility of carbamazepine in FaSSIF V1/V2/V3 and the concentration reached in the dissolution testing were in accordance with the range of the thermodynamic solubility of carbamazepine measured in HIF (range 170.0µg/mL336.0µg/mL) (Annaert et al., 2010; Clarysse et al., 2011; Heikkilä et al., 2011; Söderlind et al., 2010). Celecoxib: Celecoxib dissolution from Celebrex® 200mg capsules was far higher in FaSSIF V1, with around 12% dissolved, than in FaSSIF V2 and V3 (Fig. 2b). The first-order region at the beginning of the dissolution test revealed a significant difference between each pair of media (p<0.001). The concentrations reached in the dissolution tests of the drug product were slightly higher than the thermodynamic solubilities (48.2 ± 0.1 µg/mL vs. 41.5 ± 1 µg/mL in FaSSIF V1, 15.2 ± 0.1 µg/mL vs. 12 ± 0.2 µg/mL in FaSSIF V2 and 12 ± 0.9 µg/mL vs. 7.4 ± 0.2 µg/mL in FaSSIF V3), respectively. In the dissolution test in FaSSIF V1, the pure API achieved a similar concentration as the drug product (49.2 ± 1.1µg/mL). As no data on the solubility of celecoxib in HIF could be found in the open literature, no conclusion can be drawn as to which version of FaSSIF best reflects the in vivo solubility of celecoxib. Danazol: As for celecoxib, the dissolution of Danol® 200mg capsules was highest in FaSSIF V1 with around 5% dissolved, and considerably lower in both FaSSIF V2 and V3 (Fig. 2c). One-way ANOVA revealed a significant difference in dissolution in the first order region (p < 0.001) between each pair of media. The maximum concentrations achieved in the two-hour test were similar to the thermodynamic solubilities: 9.3 ± 0.1 µg/ml vs. 10.1 ± 0.2µg/mL in FaSSIF V1, 2.9 ± 0.1 µg/ml vs. 2.0 ± 0.0 µg/mL in FaSSIF V2 and 4.0 ± 0.1 µg/ml vs. 3.0 ± 0.1 µg/mL in FaSSIF V3. The solubilities and 7

maximum concentrations observed in the dissolution tests in all three FaSSIF versions were within the range reported in HIF (2.04µg/mL - 13.2µg/mL) (Annaert et al., 2010; Bevernage et al., 2011; Clarysse et al., 2009; Clarysse et al., 2011; Pedersen et al., 2000; Persson et al., 2005; Söderlind et al., 2010). Fenofibrate: The dissolution of fenofibrate (Cil® 160mg) was highest in FaSSIF V1, with dissolution in FaSSIF V2 and V3 less extensive but similar to each other (Fig. 2d). In the first-order region of the dissolution curve, the versions differed significantly from each other (p < 0.001). Less than 6% fenofibrate was released from the capsules within 120min in any of the FaSSIF versions. When the pure API was tested, only around 0.5% dissolved within 120 minutes in FaSSIF V1.

Maximum

concentrations achieved in the dissolution profiles of the drug product exceeded the thermodynamic solubility in each of the FaSSIF versions (18.0 ± 0.1 µg/mL vs. 11.0 ± 0.4 µg/mL in FaSSIF V1, 8.6 ± 0.1 µg/mL vs. 3.3 ± 0.2 µg/mL in FaSSIF V2, 7.5 ± 0.1 µg/mL vs. 1.4 ± 0.0 µg/mL in FaSSIF V3). In contrast, the maximum concentration reached in the dissolution test with the pure API in FaSSIF V1 was below the thermodynamic solubility (1.7 ± 0.1 µg/mL vs. 11.0 ± 0.4 µg/mL). Only the dissolution of the drug product in FaSSIF V1 was able to yield results commensurate with the thermodynamic solubility in HIF (reported range: 11.9 µg/mL - 19.7 µg/mL) (Bevernage et al., 2011; Clarysse et al., 2011). Phenytoin: Although phenytoin is a weak acid with a pKa of 8.3, it is minimally ionized over the pH range of interest (typically pH 5-7.5 in the small intestine). Phenytoin showed similar dissolution from Phenhydan® 100mg tablets in all three versions, reaching a plateau after 30 minutes (Fig. 2e). As the shape of the initial dissolution curves were similar, no statistically significant difference was detected among the versions in the first order region. On the other hand, a significant difference was observed in the plateau phase (p = 0.002). The maximum release in the different FaSSIF versions was between 19% and 23%. For FaSSIF V1, FaSSIF V2 and FaSSIF V3, the plateau concentration corresponded well to the thermodynamic solubilities (43.9 ± 0.0 µg/mL vs. 45.3 ± 0.5 µg/mL in FaSSIF V1, 38.3 ± 0.0 µg/mL vs. 41.8 ± 0.4 µg/mL in FaSSIF V2 and 46.2 ± 0.1 µg/mL vs. 41.3 ± 0.3 µg/mL in FaSSIF V3). A

8

comparison of these data for phenytoin with solubilities in HIF was not possible due to the lack of available data in the literature. 3.2.2.Weakly basic compounds Ciprofloxacin: The dissolution of ciprofloxacin hydrochloride from AL 500mg tablets initially showed a slight supersaturation in all three FaSSIF versions, reaching a maximum concentration after 10 minutes. In FaSSIF V3 the supersaturation was highest, with a maximum release of 93%, while in FaSSIF V1 and FaSSIF V2 the maximum % released was 82% and 78%, respectively. Following supersaturation, precipitation occurred in all three media (Fig. 2f), although the concentrations did not return to the thermodynamic solubility within the 120-minute duration of dissolution testing (the final concentration ranged from 668.4µg/mL - 760.5µg/mL in the dissolution test, while solubilities ranged from 183.8µg/mL - 219.5µg/mL, depending on the FaSSIF version). One-way ANOVA was applied to the precipitation phase and showed a significant difference among the versions (p = 0.035), with the post-hoc test indicating that precipitation differs only between FaSSIF V2 and FaSSIF V3. A comparison of these results with the solubility of ciprofloxacin in HIF was not possible due to the lack of available data in the literature. Dipyridamole: The dissolution of dipyridamole from Persantine® 75mg tablets was generally similar in all three versions of FaSSIF. After 90 minutes, a plateau was formed at a percentage release of between 9.8% and 11% (Fig. 2g). In the first-order region, the profiles differed significantly from each other (p = 0.003), with the difference being attributable to FaSSIF V2 according to the post-hoc test. The plateau concentration was also slightly lower in FaSSIF V2 than in FaSSIF V1 and V3. The maximum concentration achieved in the dissolution test differed only slightly from the respective thermodynamic solubility (16.6 ± 0.5 µg/mL vs. 17.4 ± 0.5 µg/mL in FaSSIF V1, 14.7 ± 0.4 µg/mL vs. 14.4 ± 0.1 µg/mL in FaSSIF V2, 16.5 ± 0.3 µg/mL vs. 15.2 ± 0.3 µg/mL in FaSSIF V3), but all concentrations were somewhat below the reported solubility of dipyridamole in HIF (20.0µg/mL29.0µg/mL) (Kalantzi et al., 2006; Kleberg et al., 2010; Söderlind et al., 2010).

9

Indinavir: The release of indinavir sulfate from Crixivan® 400mg capsules was similar in the three versions (Fig. 2h), with no significant difference according to one-way ANOVA. Like ciprofloxacin, indinavir showed supersaturation at the start of dissolution in all three FaSSIF versions, which reached a maximum after 10 minutes and ranged from 66% to 75% release. Precipitation then occurred until a plateau concentration corresponding to 7% was reached at around 60 minutes. The plateau concentrations remained well above the thermodynamic solubility in the respective media (56.4 ± 2 µg/mL vs. 28.7 ± 0.2 µg/mL in FaSSIF V1, 49.3 ± 1.1 µg/mL vs. 29.6 ± 0.4 µg/mL in FaSSIF V2, 56.4 ± 2.4 µg/mL vs. 26.1 ± 3.4 µg/mL in FaSSIF V3) and coincided with the solubility of indinavir in HIF, 51.1 µg/mL (Augustijns et al., 2014).

3.2.3.Weakly acid compounds Glibenclamide: The release of glibenclamide from Euglucon® N 3.5mg tablets reached a plateau of between 46% and 59% after 45 minutes in FaSSIF V1 and FaSSIF V2, respectively, while in FaSSIF V3 the release was over 80% in solution at the end of the test period (Fig. 2i). In both the first-order region and the plateau phase there were significant differences among the profiles (p = 0.001). Post hoc analysis revealed that the dissolution behavior responsible for the difference was in FaSSIF V3 (p<0.001). To further investigate this difference, dissolution from the glibenclamide tablets was also investigated in FaSSIF V3 adjusted to pH 6.5 (the pH of FaSSIF V1 and FaSSIF V2). This resulted in a similar dissolution profile to those in FaSSIF V1 and V2, indicating that the difference in results is attributable to the greater ionization and thus solubility of glibenclamide in FaSSIF V3, which has a pH of 6.7. The maximum concentrations in the dissolution test were marginally higher than the thermodynamic solubility in the various FaSSIF versions (3.6 ± 0.0 µg/mL vs. 3.2 ± 0.3 µg/mL in FaSSIF V1, 3.2 ± 0.0 µg/mL vs. 3.0 ± 0.2 µg/mL in FaSSIF V2, 5.7 ± 0.2 µg/mL vs. 5.7 ± 0.1 µg/mL in FaSSIF 10

V3). All these values are somewhat lower than the thermodynamic solubility in HIF, which has been reported to lie in the range 9.2µg/mL - 15.4µg/mL (Annaert et al., 2010; Clarysse et al., 2011). Ibuprofen: The dissolution of ibuprofen from Brufen® Forte 600mg tablets was complete in all three FaSSIF media and reached a plateau within 30 minutes (Fig. 2j). In the first-order region, as well as in the plateau phase, there were significant differences between the profiles (p = 0.023 and p<0.001), respectively, which could be traced back to FaSSIF V2 results in the post-hoc test. One hundred percent release was achieved in all media, corresponding to a concentration of 1.2 mg/mL, which explains why the maximum concentrations achieved were below the respective thermodynamic solubilities (1.979 ± 0.018 mg/mL in FaSSIF V1, 1.916 ± 0.032 mg/mL in FaSSIF V2 and 1.437 ± 0.015 mg/mL in FaSSIF V3). The thermodynamic solubilities in FaSSIF V1 and FaSSIF V2 correspond closely to ibuprofen’s solubility in HIF (1.99mg/mL) (Augustijns et al., 2014). Zafirlukast: The release of Zafirlukast from Accolate® 20mg tablets was similar in FaSSIF V1, FaSSIF V2, FaSSIF V3 and FaSSIF V3 pH 6.5 and ranged between 91% and 102% (Fig. 2k). In the first-order region of the dissolution profiles, the versions differed significantly from each other (p = 0.049) although no specific differences were revealed in the post-hoc test. The maximum concentration achieved in the dissolution test was between 36µg/mL and 40µg/mL, far greater than the thermodynamic solubility in FaSSIF V1 (9.2 ± 1.4 µg/mL), FaSSIF V2 (3.9 ± 0.3 µg/mL) or FaSSIF V3 (3.2 ± 0.5 µg/mL). This can be attributed to the formulation of Zafirlukast as a stabilized amorphous system. A comparison with the solubility of zafirlukast in HIF was not possible as no relevant data were found in the open literature.

Figure 2. Dissolution test profiles from a) Carbamazepine, b) Celecoxib, c) Danazol, d) Fenofibrate, e) Phenytoin, f) Ciprofloxacin, g) Dipyridamole, h) Indinavir, i) Glibenclamide, j) Ibuprofen and k) Zafirlukast in FaSSIF V1, FaSSIF V2, FaSSIF V3 and for i) and k) also in FaSSIF V3 pH 6.5 as for a), b) and d) with pure API in FaSSIF V1. Gray area indicates thermodynamic solubility in HIF.

4. Discussion 11

For the eleven compounds examined in this study, the dependency of the solubility and dissolution behavior on the version of FaSSIF vary, depending on whether the compound is not ionized over the gastro intestinal relevant pH range, is a weak base or a weak acid. Further, the alignment of these values with the solubility in HIF, was found to be compound dependent. 4.1. Nonionized compounds Regarding the nonionized compounds, in four out of five cases, the concentration of the drug products reached in the dissolution test differed from the thermodynamic solubility test value by more than 15 %, with deviations in both directions i.e. in some cases a higher concentration was reached in dissolution than in the solubility determination (ratio > 1), while in others the concentration reached in dissolution was lower than in the solubility determination (ratio < 1). The ratios of the final concentration in the dissolution test to the thermodynamic solubility for all drugs in each version of FaSSIF are listed in Table 5.

Table 5. Ratio between the final concentration in the dissolution test and thermodynamic solubility in FaSSIF V1, V2 and V3.

The origins of these differences can be manifold. In the case of carbamazepine, the thermodynamic solubility of the API was higher than the maximum value of the drug product in the dissolution test, a finding which was attributed to the shorter test duration in the dissolution test i.e. in the dissolution test there was insufficient time to reach saturation. This argument also applies to the low extent of dissolution of the pure API. Comparison of the results shows that the formulation delays the release of the API, resulting in a greater difference between thermodynamic solubility and dissolution than for the pure API. For celecoxib and fenofibrate formulations, the final concentration in the dissolution test exceeded the thermodynamic solubility. In the case of fenofibrate this can be attributed to formulation-related properties as the capsule formulation contains additional solubilizers such as sodium dodecyl sulfate and sorbitan monolaurate. The addition of solubilizers in formulations influences the surface tension and the colloidal structures in the FaSSIF versions, which 12

can lead to improvements in both wettability and solubility (Bauer et al., 2017). This was confirmed by comparing dissolution results for the drug product and pure API in FaSSIF V1, for which the concentration achieved was 10 times lower than for the drug product. Although the capsule formulation of celecoxib also contains solubilizers, this formulation has only a slight positive influence on the release. For danazol, the slight differences between thermodynamic solubility and dissolution in the three versions of FaSSIF may be attributable to batch to batch differences, since the API used for the solubility test does not originate from the same batch as the one used in the formulation. Batch to batch differences in solubility have been previously observed with danazol, as shown by the solubility studies of Bakatselou et al. and Naylor et al. (Bakatselou et al., 1991; Naylor et al., 1995). In these cases, the solubilities differed from each other, even though the same experimental set-up was used. For phenytoin the concentrations achieved in dissolution and solubility experiments were very similar (ratios were 0.97, 0.92 and 1.2 for FaSSIF V1, FaSSIF V2 and FaSSIF V3, respectively). For three of the nonionized compounds solubility data in HIF were available from the open literature. For carbamazepine and danazol, both the thermodynamic solubility and final concentration in the dissolution test in all FaSSIF versions were in accordance with the solubility in HIF. Although the thermodynamic solubility and the dissolution of danazol in the three versions were all within the range of HIF solubility, the results in FaSSIF V1 are closer to the values reported in the majority of the HIF studies (Fuchs, 2015). In the case of fenofibrate, only the dissolution in FaSSIF V1 was able to reflect the HIF solubility, since all thermodynamic solubilities, as well as the concentrations reached in dissolution in FaSSIF V2 and V3, were below the HIF solubility. From the above data, the use of biorelevant dissolution in addition to (or instead of) the thermodynamic solubility may assist in simulation of the plasma profile of prototype formulations using PBPK models. This approach enables API presentation (e.g. crystalline vs. salt, amorphous etc.) and formulation related factors influencing the in vivo dissolution to be taken into account. Although 13

the thermodynamic solubility provides a first indication of the behavior of APIs in the respective media and can be used to calculate the dissolution rate using the Noyes-Whitney equation, formulation factors can override the influence of API solubility in vivo (Dezani et al., 2013). Additionally, results for carbamazepine, danazol and fenofibrate suggest that dissolution in FaSSIF V1 shows the best correspondence with HIF solubility data. Several authors have also shown that FaSSIF V1 is able to accurately simulate the HIF, for example the review paper published by Augustijns et al., which compared solubilities in FaSSIF V1 and HIF for more than 50 active substances (Augustijns et al., 2014; Clarysse et al., 2011). Our studies likewise suggest that FaSSIF V1 yields results that are comparable with HIF solubility and that it would be the medium of choice among the three versions studied. Although Söderlind et al. found a higher correlation between solubility in FaSSIF V2 and HIF for 8 nonionized compounds, those researchers prepared the two FaSSIF versions by dissolving the individual components in phosphate buffer, rather than reconstituting the commercially available powders (Söderlind et al., 2010). Further, they only compared solubilities with one set of HIF data rather than consulting all results available in the literature. In general, the nonionized compounds investigated clearly show differences in solubility and dissolution in the FaSSIF versions. Indeed, the differences in solubility and dissolution among the FaSSIF versions tend to be higher at higher logP values (Fig. 3). In this figure, the logP values of the respective compounds are plotted against a) the ratio of the highest thermodynamic solubility measured in FaSSIF V1, V2 or V3 to the mean thermodynamic solubility in the three media or b) the highest concentration observed in dissolution of the compound in FaSSIF V1, V2 or V3 relative to the mean final dissolution concentration observed for that compound in the three media. One reason for this observed trend could be the higher proportion of non-polar areas in the colloidal structures of FaSSIF V1, enabling lyophilic substances to be better incorporated into the colloidal structures than, for example, in FaSSIF V2 (Klumpp et al., 2019). These results are also in line with the observations of Mithani et al. who demonstrated that the solubility of compounds with a log P value greater than 2.5 increases in solutions containing sodium taurocholate (Mithani et al., 1996). To further test this 14

hypothesis, more nonionized compounds would need to be investigated. The effect is somewhat ameliorated for the dissolution ratios since other factors (e.g. addition of solubilizing agents in the formulation) also come into play.

Figure 3. Ratio area (gray area) between the a) thermodynamic solubility and b) maximum dissolution in FaSSIF V1/V2/V3 versus the logP values of the nonionized compounds carbamazepine, phenytoin, celecoxib, danazol and fenofibrate.

4.2. Weak basic compounds By two of the three weak bases, the concentration reached in the dissolution test was higher than the thermodynamic solubility by more than 15% (Table 5). In these cases, the dissolution of the hydrochloride (ciprofloxacin) or sulfate (indinavir) salt, resulted in temporary supersaturation, with subsequent precipitation as the free base. But even after 120 minutes of dissolution testing, the precipitation of ciprofloxacin was not complete and thus its concentration was still far above the thermodynamic solubility at the end of the dissolution test. By contrast, indinavir precipitated more rapidly and its concentration in the dissolution medium approached the thermodynamic solubility towards the end of the experiment. Due to the fact that ciprofloxacin and indinavir were used as salts in the drug product whereas the pure API is only available as the free base, the influence of the formulation on the release was not determined. The third compound, dipyridamole, is a free base and therefore no initial supersaturation occurs in the various FaSSIF versions. Further, the ratio of concentration of dipyridamole reached in the dissolution test to its thermodynamic solubility was similar in all three media. For two of the weak bases solubility data in HIF were available. For indinavir the thermodynamic solubility in the FaSSIF versions underestimates the actual HIF solubility, whereas the final concentrations in the dissolution tests were all close to the HIF solubility measured by Augustjins et al. (Augustijns et al., 2014). In the case of dipyridamole, the thermodynamic solubility underestimates the HIF solubility, whereas the dissolution results approach the lower bound of 15

solubility in HIF. Additional two-stage dissolution tests with dipyridamole in FaSSGF and FaSSIF V1/V2 have shown that precipitation occurs after transfer into the small intestine medium, with concentrations of dipyridamole after 2h of about 40µg/mL and thus somewhat higher than the solubility range in HIF (data not shown). These results suggest that, for weakly basic compounds, results from dissolution testing in FaSSIF may be helpful to forecast the in vivo behavior of salt forms, whereas a two stage test might be required for free bases. Additionally, physiologically based pharmacokinetic (PBPK) model could help to translate the solubility, dissolution and transfer data to the in vivo performance (Kaur et al., 2018). For the weakly bases, it can be seen that the differences between the FaSSIF versions, with respect to the thermodynamic solubilities as well as the final dissolution concentrations, are minimal. Another study comparing the solubility of ionized compounds between FaSSIF V1 and V2 came to a similar conclusion (Söderlind et al., 2010).

4.3. Weak acidic compounds For two of the three weak acids, the concentration reached in the dissolution test differed from the thermodynamic solubility by more than 15% (Table 5). In the case of ibuprofen, the final concentration was lower than the thermodynamic solubility simply because the dose used in the dissolution test was insufficient to achieve the saturation solubility. In order to investigate the influence of the formulation of ibuprofen on the dissolution, a higher dose strength or multiple dosage units would have to be used. For zafirlukast a formulation effect occurred, resulting in a higher concentration in the dissolution test than the thermodynamic solubility: in this case the API in the formulation is present in the amorphous form and additional stabilizers are present (Madsen et al., 2016). Since the pure API, in contrast to the drug product, does not exist in an amorphous state, the influence of the formulation could not be directly investigated. For glibenclamide the ratios of 16

the concentrations achieved in the dissolution tests and the thermodynamic solubilities were within a 15% deviation. Of the weak acids, solubility data in HIF were only available for glibenclamide. Both the thermodynamic solubility and the final dissolution concentration underestimated the HIF solubility by a factor of about two. The fact that the dose used in the dissolution test was too low to achieve solubility in HIF is of no consequence as the thermodynamic solubility in the FaSSIF versions also did not achieve the HIF solubility range. The solubilities of glibenclamide in the respective HIF sample cannot be assigned on the basis of the data available, but it can be seen that the pH values of the samples exceed the pH values of the FaSSIF versions (1: geometric mean 6.7 (range: 5.41-7.17); 2: range 6.5-7.3) (Annaert et al., 2010; Clarysse et al., 2011). In this respect, it is possible that the higher solubility of glibenclamide in HIF resulted from the higher HIF pH value. Of the three FaSSIF versions, both the solubility and the dissolution in FaSSIF V3 were closest to the HIF solubility. This can be attributed to the 0.2-unit higher pH value in FaSSIF V3 compared to FaSSIF V1 and V2, since the solubility in FaSSIF V3 pH 6.5 was very similar to the solubilities in FaSSIF V1 and V2. Another study comparing the solubility of ionized compounds between FaSSIF V1 and V2 came to a similar conclusion (Söderlind et al., 2010). In general, leaving aside the pH difference between FaSSIF V3 and the other versions, the more complex composition of FaSSIF V3 seems to have little effect on solubility or dissolution. 5. Conclusion The different compositions of the FaSSIF versions can have an influence on the solubility and dissolution, and the effect is most pronounced for nonionized compounds. In the case of weakly acidic compounds, the pH value plays a major role. Dissolution tests are an important additional method for the evaluation of FaSSIF versions, as they are superior to describe the properties of weakly bases and additionally consider formulation-related influences. The investigations suggest that FaSSIF V1 is the most consistent predictor of simulating small intestine fluids. 17

6. Funding This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

References 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 intestinal fluids support age-dependent oral drug absorption. European journal of pharmaceutical sciences : official journal of the European Federation for Pharmaceutical Sciences 39 (1-3), 15–22, https://doi.org/10.1016/j.ejps.2009.10.005. Augustijns, P., Wuyts, B., Hens, B., Annaert, P., Butler, J., Brouwers, J., 2014. A review of drug solubility in human intestinal fluids: Implications for the prediction of oral absorption. European journal of pharmaceutical sciences : official journal of the European Federation for Pharmaceutical Sciences 57, 322–332, https://doi.org/10.1016/j.ejps.2013.08.027. Bakatselou, V., Oppenheim, R.C., Dressman, J.B., 1991. Solubilization and Wetting Effects of Bile Salts on the Dissolution of Steroids. Pharmaceutical research 08 (12), 1461–1469, https://doi.org/10.1023/A:1015877929381. Bauer, K.H., Frömming, K.-H., Führer, C., Lippold, B.C., Müller-Goymann, C., Schubert, R., Breitkreutz, J., 2017. Pharmazeutische Technologie: Mit Einführung in Biopharmazie und Biotechnologie : mit 291 Abbildungen und 92 Tabellen, 10., überarbeitete und aktualisierte Auflage ed. Wissenschaftliche Verlagsgesellschaft, Stuttgart, XXVI, 824 Seiten. Berben, P., Ashworth, L., Beato, S., Bevernage, J., Bruel, J.-L., Butler, J., Dressman, J., Schäfer, K., Hutchins, P., Klumpp, L., Mann, J., Nicolai, J., Ojala, K., Patel, S., Powell, S., Rosenblatt, K., Tomaszewska, I., Williams, J., Augustijns, P., 2019. Biorelevant dissolution testing of a weak base: Interlaboratory reproducibility and investigation of parameters controlling in vitro precipitation. European journal of pharmaceutics and biopharmaceutics : official journal of Arbeitsgemeinschaft fur Pharmazeutische Verfahrenstechnik e.V 140, 141–148, https://doi.org/10.1016/j.ejpb.2019.04.017. Bevernage, J., Forier, T., Brouwers, J., Tack, J., Annaert, P., Augustijns, P., 2011. Excipient-mediated supersaturation stabilization in human intestinal fluids. Molecular pharmaceutics 8 (2), 564–570, https://doi.org/10.1021/mp100377m. Biorelevant.com. How to make FaSSIF/FeSSIF/FaSSGF and FaSSIF V2. https://biorelevant.com/fassiffessif-fassgf/make/. Accessed 8 March 2018. Clarysse, S., Brouwers, J., Tack, J., Annaert, P., Augustijns, P., 2011. Intestinal drug solubility estimation based on simulated intestinal fluids: Comparison with solubility in human intestinal fluids. European journal of pharmaceutical sciences : official journal of the European Federation for Pharmaceutical Sciences 43 (4), 260–269, https://doi.org/10.1016/j.ejps.2011.04.016. Clarysse, S., Psachoulias, D., Brouwers, J., Tack, J., Annaert, P., Duchateau, G., Reppas, C., Augustijns, P., 2009. Postprandial changes in solubilizing capacity of human intestinal fluids for BCS class II drugs. Pharmaceutical research 26 (6), 1456–1466, https://doi.org/10.1007/s11095-009-9857-7. Dezani, A.B., Pereira, T.M., Caffaro, A.M., Reis, J.M., Serra, C.H.d.R., 2013. Equilibrium solubility versus intrinsic dissolution: characterization of lamivudine, stavudine and zidovudine for BCS classification. Braz. J. Pharm. Sci. 49 (4), 853–863, https://doi.org/10.1590/S198482502013000400026. 18

Dhabu, P.M., Akamanchi, K.G., 2002. A stability-indicating HPLC method to determine Celecoxib in capsule formulations. Drug development and industrial pharmacy 28 (7), 815–821, https://doi.org/10.1081/DDC-120005627. Dressman, J.B., Amidon, G.L., Reppas, C., Shah, V.P., Dressman, J.B., Amidon, G.L., Reppas, C., Shah, V.P., 1998. Dissolution testing as a prognostic tool for oral drug absorption: immediate release dosage forms. Pharmaceutical research 15 (1), 11–22, https://doi.org/10.1023/A:1011984216775. EMEA, 2005. Scientific Discussion - Crixivan. https://www.ema.europa.eu/en/documents/scientificdiscussion/crixivan-epar-scientific-discussion_en.pdf. Accessed 11 March 2019. Fuchs, A., 2015. FaSSIF-V3: Entwicklung einer neuen Generation von Biorelevanten Medien zur Simulation des nüchternen humanen Dünndarms. Dissertation, Frankfurt am Main. Fuchs, A., Leigh, M., Kloefer, B., Dressman, J.B., 2015. Advances in the design of fasted state simulating intestinal fluids: FaSSIF-V3. European journal of pharmaceutics and biopharmaceutics : official journal of Arbeitsgemeinschaft fur Pharmazeutische Verfahrenstechnik e.V 94, 229–240, https://doi.org/10.1016/j.ejpb.2015.05.015. Galia, E., Nicolaides, E., Horter, D., Lobenberg, R., Reppas, C., Dressman, J.B., 1998. Evaluation of various dissolution media for predicting in vivo performance of class I and II drugs. Pharmaceutical research 15 (5), 698–705, https://doi.org/10.1023/A:1011910801212. Gamsiz, E.D., Ashtikar, M., Crison, J., Woltosz, W., Bolger, M.B., Carrier, R.L., 2010. Predicting the effect of fed-state intestinal contents on drug dissolution. Pharmaceutical research 27 (12), 2646–2656, https://doi.org/10.1007/s11095-010-0264-x. Glomme, A., März, J., Dressman, J.B., 2005. Comparison of a miniaturized shake-flask solubility method with automated potentiometric acid/base titrations and calculated solubilities. Journal of Pharmaceutical Sciences 94 (1), 1–16, https://doi.org/10.1002/jps.20212. Greupink, R., Schreurs, M., Benne, M.S., Huisman, M.T., Russel, F.G.M., 2013. Semi-mechanistic physiologically-based pharmacokinetic modeling of clinical glibenclamide pharmacokinetics and drug-drug-interactions. European journal of pharmaceutical sciences : official journal of the European Federation for Pharmaceutical Sciences 49 (5), 819–828, https://doi.org/10.1016/j.ejps.2013.06.009. Hansmann, S., Miyaji, Y., Dressman, J., 2018. An in silico approach to determine challenges in the bioavailability of ciprofloxacin, a poorly soluble weak base with borderline solubility and permeability characteristics. European journal of pharmaceutics and biopharmaceutics : official journal of Arbeitsgemeinschaft fur Pharmazeutische Verfahrenstechnik e.V 122, 186–196, https://doi.org/10.1016/j.ejpb.2017.10.019. Heikkilä, T., Karjalainen, M., Ojala, K., Partola, K., Lammert, F., Augustijns, P., Urtti, A., Yliperttula, M., Peltonen, L., Hirvonen, J., 2011. Equilibrium drug solubility measurements in 96-well plates reveal similar drug solubilities in phosphate buffer pH 6.8 and human intestinal fluid. International journal of pharmaceutics 405 (1-2), 132–136, https://doi.org/10.1016/j.ijpharm.2010.12.007. Jantratid, E., Janssen, N., Reppas, C., Dressman, J.B., 2008. Dissolution media simulating conditions in the proximal human gastrointestinal tract: an update. Pharmaceutical research 25 (7), 1663– 1676, https://doi.org/10.1007/s11095-008-9569-4. Kalantzi, L., Persson, E., Polentarutti, B., Abrahamsson, B., Goumas, K., Dressman, J.B., Reppas, C., 2006. Canine intestinal contents vs. simulated media for the assessment of solubility of two weak bases in the human small intestinal contents. Pharmaceutical research 23 (6), 1373–1381, https://doi.org/10.1007/s11095-006-0207-8. Kaur, N., Narang, A., Bansal, A.K., 2018. Use of biorelevant dissolution and PBPK modeling to predict oral drug absorption. European journal of pharmaceutics and biopharmaceutics : official journal

19

of Arbeitsgemeinschaft fur Pharmazeutische Verfahrenstechnik e.V 129, 222–246, https://doi.org/10.1016/j.ejpb.2018.05.024. Kleberg, K., Jacobsen, J., Müllertz, A., 2010. Characterising the behaviour of poorly water soluble drugs in the intestine: Application of biorelevant media for solubility, dissolution and transport studies. The Journal of pharmacy and pharmacology 62 (11), 1656–1668, https://doi.org/10.1111/j.2042-7158.2010.01023.x. Klumpp, L., Nagasekar, K., McCullough, O., Seybert, A., Ashtikar, M., Dressman, J., 2019. Stability of Biorelevant Media Under Various Storage Conditions. Dissolution Technol. 26 (26 // 2), 6–18, https://doi.org/10.14227/DT260219P6. Lobenberg, R., Kramer, J., Shah, V.P., Amidon, G.L., Dressman, J.B., Löbenberg, R., Krämer, J., 2000. Dissolution testing as a prognostic tool for oral drug absorption: dissolution behavior of glibenclamide. Pharmaceutical research 17 (4), 439–444, https://doi.org/10.1023/A:1007529020774. Madsen, C.M., Boyd, B., Rades, T., Müllertz, A., 2016. Supersaturation of zafirlukast in fasted and fed state intestinal media with and without precipitation inhibitors. European journal of pharmaceutical sciences : official journal of the European Federation for Pharmaceutical Sciences 91, 31–39, https://doi.org/10.1016/j.ejps.2016.05.026. Mann, J., Dressman, J., Rosenblatt, K., Ashworth, L., Muenster, U., Frank, K., Hutchins, P., Williams, J., Klumpp, L., Wielockx, K., Berben, P., Augustijns, P., Holm, R., Hofmann, M., Patel, S., Beato, S., Ojala, K., Tomaszewska, I., Bruel, J.-L., Butler, J., 2017. Validation of Dissolution Testing with Biorelevant Media: An OrBiTo Study. Molecular pharmaceutics 14 (12), 4192–4201, https://doi.org/10.1021/acs.molpharmaceut.7b00198. Markopoulos, C., Andreas, C.J., Vertzoni, M., Dressman, J., Reppas, C., 2015. In-vitro simulation of luminal conditions for evaluation of performance of oral drug products: Choosing the appropriate test media. European journal of pharmaceutics and biopharmaceutics : official journal of Arbeitsgemeinschaft fur Pharmazeutische Verfahrenstechnik e.V 93, 173–182, https://doi.org/10.1016/j.ejpb.2015.03.009. Mithani, S.D., Bakatselou, V., TenHoor, C.N., Dressman, J.B., 1996. Estimation of the Increase in Solubility of Drugs as a Function of Bile Salt Concentration // Estimation of the increase in solubility of drugs as a function of bile salt concentration. Pharmaceutical research 13 (1), 163– 167, https://doi.org/10.1023/a:1016062224568. 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. Molecular pharmaceutics 11 (8), 2835–2844, https://doi.org/10.1021/mp500196f. Nair, P.P., Kritchevsky, D., 1971. The Bile Acids Chemistry, Physiology, and Metabolism: Volume 1: Chemistry. Springer US; Imprint; Springer, Boston, MA. Naylor, L.J., Bakatselou, V., Rodríguez-Hornedo, N., Weiner, N.D., Dressman, J.B., 1995. Dissolution of steroids in bile salt solutions is modified by the presence of lecithin. Eur. J. Pharm. Biopharm 1995 (41), 346–353. Otsuka, K., Shono, Y., Dressman, J., 2013. Coupling biorelevant dissolution methods with physiologically based pharmacokinetic modelling to forecast in-vivo performance of solid oral dosage forms. The Journal of pharmacy and pharmacology 65 (7), 937–952, https://doi.org/10.1111/jphp.12059. Pedersen, B.L., Müllertz, A., Brøndsted, H., Kristensen, H.G., 2000. A comparison of the solubility of danazol in human and simulated gastrointestinal fluids. Pharmaceutical research 17 (7), 891–894, https://doi.org/10.1023/A:1007576713216. Persson, E.M., Gustafsson, A.-S., Carlsson, A.S., Nilsson, R.G., Knutson, L., Forsell, P., Hanisch, G., Lennernäs, H., Abrahamsson, B., 2005. The effects of food on the dissolution of poorly soluble 20

drugs in human and in model small intestinal fluids. Pharmaceutical research 22 (12), 2141–2151, https://doi.org/10.1007/s11095-005-8192-x. Pubchem. Celecoxib. https://pubchem.ncbi.nlm.nih.gov/compound/celecoxib. Accessed 11 March 2019. Pubchem. Fenofibrate. https://pubchem.ncbi.nlm.nih.gov/compound/fenofibrate. Accessed 11 March 2019. Pubchem. Ibuprofen. https://pubchem.ncbi.nlm.nih.gov/compound/ibuprofen. Accessed 11 March 2019. Pubchem. Phenytoin. https://pubchem.ncbi.nlm.nih.gov/compound/1775. Accessed 11 March 2019. Pubchem. Zafirlukast. https://pubchem.ncbi.nlm.nih.gov/compound/5717. Accessed 11 March 2019. Pubchem Carbamazepine. Carbamazepine. https://pubchem.ncbi.nlm.nih.gov/compound/carbamazepine. Accessed 11 March 2019. Rajadhyaksha, N.S., Jain, S.P., Amin, P.D., 2007. Carbamazepine: Stability Indicating HPLC Assay Method. Analytical Letters 40 (13), 2506–2514, https://doi.org/10.1080/00032710701583557. Shah, V.P., Ogger, K.E., 1986. Comparison of Ultraviolet and Liquid Chromatographic Methods for Dissolution Testing of Sodium Phenytoin Capsules. Journal of Pharmaceutical Sciences 75 (11), 1113–1115, https://doi.org/10.1002/jps.2600751119. Söderlind, E., Karlsson, E., Carlsson, A., Kong, R., Lenz, A., Lindborg, S., Sheng, J.J., 2010. Simulating fasted human intestinal fluids: Understanding the roles of lecithin and bile acids. Molecular pharmaceutics 7 (5), 1498–1507, https://doi.org/10.1021/mp100144v. Varma, M.V.S., Scialis, R.J., Lin, J., Bi, Y.-A., Rotter, C.J., Goosen, T.C., Yang, X., 2014. Mechanismbased pharmacokinetic modeling to evaluate transporter-enzyme interplay in drug interactions and pharmacogenetics of glyburide. The AAPS journal 16 (4), 736–748, https://doi.org/10.1208/s12248-014-9614-7. Wu, S.-S., Chein, C.-Y., Wen, Y.-H., 2008. Analysis of ciprofloxacin by a simple high-performance liquid chromatography method. Journal of chromatographic science 46 (6), 490–495, https://doi.org/10.1093/chromsci/46.6.490.

Tables and Figures: Graphical Abstract:

Tables:

Cholesterol (mM)

FaSSGF -

FaSSIF V1 -

FaSSIF V2 -

FaSSIF V3 0.2

21

Lecithin (mM) Lysolecithin (mM) Sodium glycocholate (mM) Sodium oleat (mM) Sodium taurocholate (mM) Hydrochloric acid Maleic acid (mM) Potassium dihydrogen phosphate (mM) Sodium chloride (mM) Sodium hydroxide (mM) pH

0.02 0.08 qs pH 1.6

34.2 1.6

0.75 3

0.2 3

0.035 0.315 1.4 0.315 1.4

28.65 105.85 10.5 6.5

19.12 68.62 34.8 6.5

10.26 93.3 16.56 6.7 / 6.5*

Table 1. Compositions of biorelevant media used for solubility and dissolution studies * The pH of FaSSIF V3 was decreased to a value of 6.5 for certain experiments.

Carbamazepine Celecoxib

Commercial formulations studied Carbamazepin 200 - 1a Pharma® Celebrex® 200mg

Danazol

Danol® 100mg Capsules

Fenofibrate Phenytoin Ciprofloxacin hydrochloride

Cil® 160mg Phenhydan® 100mg

Sanofi, UK Wintrop Arzneimittel GmbH, Germany Destin Arzneimittel GmbH, Germany

Ciprofloxacin AL 500mg

Aluid Pharma® GmbH, Germany

Tablet

Weak base

Dipyridamole

Persantine® 75mg

Boehringer Ingelheim, France

Tablet

Weak base

Indinavir sulfate

Crixivan® 400mg

Capsule

Weak base

Glibenclamide Ibuprofen

Euglucon ® N 3.5mg Brufen® Forte 600mg

Merck Sharp & Dhome Limited, UK Sanofi-Aventis Deutschland GmbH, Germany Mylan EPD SPRL-BVBA, Belgium

Tablet Tablet

Weak acid Weak acid

Accolate® 20mg

AstraZeneca AG, Switzerland

Tablet

Weak acid

Compound

Zafirlukast

Producer

Formulation

Classification

pKa

log P 2.5² (Pubchem Carbamazepine) 3.5² (Pubchem) 4.5² (Bakatselou et al., 1991)

1A Pharma GmbH, Germany Pfizer GmbH, Germany

Tablet Capsule

Nonionized Nonionized

/ /

Capsule

Nonionized

/

Capsule Tablet

Nonionized 1 Nonionized

/ / 8.6 (Hansmann et al., 2018) 6.2 (Nair and Kritchevsky, 1971) 3.8 / 6.2 (EMEA, 2005) 5.3 (Greupink et al., 2013) 5.3 (Pubchem) ~4 (Madsen et al., 2016)

5.3 (Pubchem) 2.5² (Pubchem) -0.7² (Hansmann et al., 2018) 1.4² (Gamsiz et al., 2010) 2.7 (EMEA, 2005) 3.1² (Varma et al., 2014) 4.0 (Pubchem) 5.4 (Pubchem)

Table 2. APIs investigated, with details of the commercial product tested, manufacturer, formulation type, classification by ionization, pKa and log P value. 1 Phenytoin is not ionized over the pH range studied (pH 6.5 – 6.7) ²Values has been rounded to two significant digits Mobile Phase

pH

Carbamazepine

Water/Acetonitrile/Methanol (35/40/25)

-

Celecoxib

Methanol/Water/Phosphoric acid (75/25)

3.5

Danazol

Acetonitrile/Water (70/30)

-

Fenofibrate

Acetonitrile/Water/Phosphoric acid (80/20)

2.5

Phenytoin Ciprofloxacin hydrochloride

Methanol/Water (55/45) Phosphoric acid 0.025M/Acetornitrile/ Triethylamine (86/14)

3

Dipyridamole

Methanol/Phosphate buffer (4/1)

4.6

Indinavir sulfate

Citric acid buffer/ ACN (3/2)

4.9

Glibenclamide

4

Ibuprofen

Phosphate buffer 0.025M/ACN (55/45) Acetonitrile/Water/Triethylamine/Phosphoric acid (600/400/2.65/1.4)

Zafirlukast

Acetonitrile/Water/Phosphoric acid (600/400)

3

3

Column LiChroCART® 250-4.6, RP-18e, 5µm LiChroCART® 125-4.6, RP-18e, 5µm Purospher® Star 1504.6, RP-18e, 5µm LiChroCART® 125-4.6, RP-18, 5µm ZORBAX Eclipse XDBC18, 4.6x250mm, 5µm Purospher® Star 1504.6, RP-18e, 5µm LiChroCART® 150-4.6, RP-18e, 5µm ZORBAX RX-C8, 150x5mm, 5µm LiChroCART® 150-4.6, RP-18e, 5µm Kinetex XB-C18 100A, 100x4.6mm, 2.6µm Spherisorb ODS-2 C18, 50x4.6mm, 5µm

Wavelength (nM)

Flow rate (mL/min)

Injection volume (µl)

LOQ (µg/mL)

286

1.2

10

0.06

252

0.8

20

0.98

286

1.2

50

0.05

254

1

50

0.05

220

1.5

50

0.747

278

1.3

10

0.11

288

1.5

50

0.14

260

1.2

10

1.44

230

1.4

50

0.13

254

1.1

10

4.86

224

1.5

10

0.47

Table 3. HPLC setup for the analysis of each compound

22

API

Ionization

FaSSIF V1 (µg/mL)

FaSSIF V2 (µg/mL)

FaSSIF V3 (µg/mL)

HIF (µg/mL)

Carbamazepine*

Nonionized

298.1 ± 15.9

295.3 ± 16.8

311.7 ± 2.5

170.0 – 336.0

Celecoxib*

Nonionized

41.5 ± 1.0

12.0 ± 0.2

7.4 ± 0.2

/

Danazol*

Nonionized

10.1 ± 0.2

2.0 ± 0.0

3.0 ± 0.1

2.04 – 13.2

Fenofibrate*

Nonionized

11.0 ± 0.4

3.3 ± 0.2

1.4 ± 0.0

11.9 – 19.7

Phenytoin

Nonionized

45.3 ± 0.5

41.8 ± 0.4

41.3 ± 0.3

/

Ciprofloxacin hydrochloride

Weak base

219.55 ± 6.45

183.82 ± 6.26

196.77 ± 3.39

/

Dipyridamole

Weak base

17.4 ± 0.5

14.4 ± 0.1

15.2 ± 0.3

20.0 – 29.0

Indinavir sulfate

Weak base

28.7 ± 0.2

29.6 ± 0.4

26.1 ± 3.4

51.1

Glibenclamide*

Weak acid

3.2 ± 0.3

3.0 ± 0.2

5.7 ± 0.1

9.2 – 15.4

Ibuprofen

Weak acid

1979.8 ± 18.34

1916.9 ± 32.1

1437.3 ± 14.9

1990

Zafirlukast

Weak acid

9.2 ± 1.4

3.9 ± 0.3

3.2 ± 0.5

/

Table 4. Solubility (µg/mL) of 11 APIs in FaSSIF V1, V2, V3 and HIF. Solubility of APIs* had been measured by Fuchs et al.(Fuchs, 2015). References for HIF solubilities are given in the discussion. Log P values of the different compounds can be found in table 2.

API Carbamazepine Celecoxib Danazol Fenofibrate Phenytoin Ciprofloxacin hydrochloride Dipyridamole Indinavir sulfate Glibenclamide Zafirlukast

Ionization Nonionized Nonionized Nonionized Nonionized Nonionized Weak base Weak base Weak base Weak acid Weak acid

FaSSIF V1 0.74 1.16 0.92 1.64 0.97 3.21 0.96 1.97 1.14 4.16

FaSSIF V2 0.78 1.27 1.45 2.61 0.92 3.64 1.02 1.66 1.08 9.47

FaSSIF V3 0.65 1.62 1.33 5.34 1.12 3.87 1.09 2.16 1.01 12.38

Table 5. Ratio between the final concentration in the dissolution test and thermodynamic solubility in FaSSIF V1, V2 and V3.

Figures:

4

% dissolved

3

2

1

0 0

20

40

60

80

100

120

Time (min) Plateau First Order Model

Figure 1. Fitting of the dissolution results. The profile which was divided into 2 regions for statistical analysis, a first order region (short lines) and, where appropriate, a plateau region (straight line). The example shown is for Celecoxib tablets.

23

a) Carbamazepine Dissolved in % and mg/mL

b) Celecoxib Dissolved in % and mg/mL 0,35

14

0,30

12

0,25

10

0,05

40

0,15

0,04

8

0,03

6 0,02

0,10

4

0,05

2

0,00

0

mg/mL dissolved

0,20

% dissolved

% dissolved

60

mg/mL dissolved

80

20

0 0

20

40

60

80

100

0,01

120

0,00 0

20

40

Time (min)

60

80

100

120

Time (min)

FaSSIF V1 FaSSIF V2 FaSSIF V3 FaSSIF V1 Pure API

FaSSIF V1 FaSSIF V2 FaSSIF V3 FaSSIF V1 Pure API

c) Danazol Dissolved in % and mg/mL 0,014

6

0,012

4

0,008

0,006

2

mg/mL dissolved

% dissolved

0,010

0,004

0,002

0

0,000 0

20

40

60

80

100

120

Time (min) FaSSIF V1 FaSSIF V2 FaSSIF V3

24

e) Phenytoin Dissolved in % and mg/mL

d) Fenofibrate Dissolved in % and mg/mL

0,05

20

0,04

15

0,03

10

0,02

5

0,01

mg/mL dissolved

0,010

% dissolved

0,015 4

mg/mL dissolved

0,020

6

% dissolved

25

2 0,005

0

0,000 0

20

40

60

80

100

0

120

0,00 0

Time (min)

20

40

60

80

100

120

Time (min)

FaSSIF V1 FaSSIF V2 FaSSIF V3 FaSSIF V1 Pure API

FaSSIF V1 FaSSIF V2 FaSSIF V3

100

1,0

80

0,8

60

0,6

40

0,4

20

0,2

0

mg/mL dissolved

% dissolved

f) Ciprofloxacin Dissolved in % and mg/mL

0,0 0

20

40

60

80

100

120

Time (min) FaSSIF V1 FaSSIF V2 FaSSIF V3

25

g) Dipyridamole Dissolved in % and mg/mL 0,030

20

0,025

% dissolved

0,020

10

0,015

0,010

mg/mL dissolved

15

5 0,005

0

0,000 0

20

40

60

80

100

120

Time (min) FaSSIF V1 FaSSIF V2 FaSSIF V3

h) Indinavir Dissolved in % and mg/mL

80

% dissolved

60 0,4 40

mg/mL dissolved

0,6

0,2 20

0

0,0 0

20

40

60

80

100

120

Time (min) FaSSIF V1 FaSSIF V2 FaSSIF V3

j) Ibuprofen Dissolved in % and mg/mL

i) Glibenclamide Dissolved in % and mg/mL 0,016

120

0,006

% dissolved

0,008 100

mg/mL dissolved

0,010

1,2 1,0

80

0,8 60 0,6

mg/mL dissolved

100

0,012 150

% dissolved

1,4

0,014

200

40 0,004

50

0,002 0

0,000 0

20

40

60

80

100

120

0,4 20

0,2

0

0,0 0

20

Time (min) FaSSIF V1 FaSSIF V2 FaSSIF V3 FaSSIF V3 pH 6.5

40

60

80

100

120

Time (min) FaSSIF V1 FaSSIF V2 FaSSIF V3

26

k) Zafirlukast Dissolved in % and mg/mL 120

0,04

% dissolved

80 0,03 60 0,02 40

mg/mL dissolved

100

0,01 20

0

0,00 0

20

40

60

80

100

120

Time (min) FaSSIF V1 FaSSIF V2 FaSSIF V3 FaSSIF V3 pH 6.5

Figure 2. Dissolution test profiles from a) Carbamazepine, b) Celecoxib, c) Danazol, d) Fenofibrate, e) Phenytoin, f) Ciprofloxacin, g) Dipyridamole, h) Indinavir, i) Glibenclamide, j) Ibuprofen and k) Zafirlukast in FaSSIF V1, FaSSIF V2, FaSSIF V3 and for i) and k) also in FaSSIF V3 pH 6.5, as for a), b) and d) with pure API in FaSSIF V1. Gray area indicates thermodynamic solubility in HIF. b) Influence of logP on final dissolution concentration ratio

Ratio between final dissolution concentration in FaSSIF V1/V2/V3

Ratio between thermodynamic solubility in FaSSIF V1/V2/V3

a) Influence of logP on thermodynamic solubility ratio

2,0

1,5

1,0

0,5

2,0

1,5

1,0

0,5

0,0

Carbamazepin 2.45

Phenytoin 2.47

Celecoxib 3.53

Danazol 4.53

Fenofibrat 5.3

Carbamazepin 2.45

Phenytoin 2.47

Celecoxib 3.53

Danazol 4.53

Fenofibrat 5.3

Figure 3. Ratio area (gray area) between the a) thermodynamic solubility and b) maximum dissolution in FaSSIF V1/V2/V3 versus the logP values of the nonionized compounds carbamazepine, phenytoin, celecoxib, danazol and fenofibrate.

27