Two-Stage Single-Compartment Models to Evaluate Dissolution in the Lower Intestine

Materials

Sulfasalazine powder (98.1% purity) was from Fluka Analytical (Buchs, Switzerland), and Azulfidine® immediate release tablets (500 mg/tab; Pfizer, Berlin, Germany) were purchased locally.

L-870,810 anhydrous crystalline powder (99.9% purity), granules of L-870,810 containing 52.5% (w/w) L-870,810 (50%, w/w free acid), and hard gelatine capsules (for testing the granules) were provided by Merck Research Laboratories (Merck& Co., Inc., Kenilworth, NJ, USA). L-870,810 exhibits good chemical stability in the solid state. Granules of L-870,810 were comprised of standard non-functional excipients (standard sugars as fillers along with disintegrants, lubricants, and binders) with no surfactant incorporated. None of these excipients would be expected to affect drug solubility. Protection from light is necessary for long-term storage of L-870,810 (Merck data on file). Solutions were protected from light and stability of the API was confirmed for all procedures applied in the present investigation.

Mesalamine powder (≥99% purity) was from Sigma–Aldrich (Saint Louis, Missouri) and Asacol® tablets (400 mg/tab; Tillots Pharma AG, Switzerland) were purchased locally.

Egg phosphatidylcholine (Lipoid E PC® 99.1% purity, lot# 105019-1/14) was kindly donated by Lipoid GmbH (Ludwigshafen, Germany). SIF® Powder Original was kindly donated by biorelevant.com (Surrey, United Kingdom). All other reagents were of analytical or HPLC grade and purchased commercially.

Evaluation of Sulfasalazine Solubility in Intestinal Lumen and of Azulfidine® Tablets Dissolution in Upper Intestinal Lumen

Two-Stage Single-Compartment Models for Evaluating Dissolution in Distal Ileum and Proximal Colon

Dissolution was studied in a single-compartment setup using a medium simulating the environment in the distal ileum (Stage 1, 0–2 h) and a medium simulating the environment in the ascending colon (Stage 2, 2–6 h). Although retention of non-disintegrating dosage forms at the ileocecal valve can occur, especially in the fasted state, residence of micropellets in distal ileum is generally shorter than 2 h.21, 22 In the present study, experiments lasted for 2 h to ensure adequate characterization of dissolution profile. Similarly, duration of experiments under conditions simulating the environment in the ascending colon was based on the average residence time in the region23 and the need for collecting adequate information for characterizing the dissolution process.

Change of composition of the medium simulating the conditions in the distal ileum (SIF_ileum) to a medium simulating the conditions in ascending colon in the fasted state (FaSSCoF) and in the fed state (FeSSCoF) was achieved by adding an appropriate solution at the end of Stage 1. The same medium simulating the conditions in the distal ileum (SIF_ileum) was used in both fasted and fed state, as recent data indicate that conditions at distal ileum are not affected significantly by dosing conditions.5 The ratio of volumes in Stage 1 and Stage 2 was 1:5, that is, similar to the average ratio of aqueous volumes in distal ileum and ascending colon of healthy adults.3, 5 Volumes used during Stage 1 and during Stage 2 were a compromise between the actual ratio of intraluminal volumes and practicalities, for example, ability to test various types of formulations and draw multiple samples (Tables 1 and 2). Compositions of SIF_ileum, of the added solutions, and of the resulting solutions (FaSSCoF and FeSSCoF) are presented in Table 1 (when fasted state conditions are to be simulated) and in Table 2 (when fed state conditions are to be simulated). Compositions were based on available literature data and assuming two levels of simulation of the luminal environment.7 In the first level of simulation (Level I), only the pH and buffer capacity was simulated. In Level II simulation, bile components, digestion products, and osmolality, are additionally simulated (Tables 1 and 2).

The impact of type/intensity of fluid convection was evaluated by using the mini-paddle apparatus with the paddle rotating at 100 rpm (Fig. 1) and the compendial flow-through apparatus (closed-loop mode) operating at 4 mL/min (Fig. 2).

To simulate dosing conditions previously applied in human studies13, 24, 25 and the presence of undissolved API in the lower intestine as much as possible, experiments were performed by using one Azulfidine® tablet under simulated fasted and fed state conditions (500 mg sulfasalazine), L-870,810 granules under simulated fasted state conditions [equivalent to 400 mg of API (free acid) placed in two hard gelatin capsules] and three Asacol® tablets under simulated fasted and fed state conditions (1200 mg mesalamine). It may be argued that after administration of three Asacol® tablets, transit through the upper GI lumen may vary among the three tablets. However, in the majority of cases, orally administered monolithic dosage forms accumulate at the ileocecal valve prior to colon arrival.26

Experiments with the mini-paddle apparatus were performed using a Distek dissolution apparatus in conjunction with a model 2100β dissolution system (Distek, New Brunswick, New Jersey). The lower end of the paddle during Stage 1 (volume of medium: 40 mL) was 1 cm from the bottom of the mini vessel and sampling was performed from the middle of the distance between the top of the paddle and the surface of the medium (Fig. 1). In order to achieve better mixing of the contents of mini vessel, during Stage 2 (volume of medium: 200 mL), the paddle was raised by 2 cm and sampling was performed from the middle of the distance between the top of the paddle and the surface of the medium (Fig. 1).

Experiments with the flow-through apparatus were performed with an Erweka flow-through dissolution apparatus (Erweka DFZ60, Heusenstamm, Germany) equipped with a piston pump (HKP60, Heusenstamm, Germany). A 5 mm-sized glass bead was placed in the tip of each dissolution cell and a total of 1.7 g of 1 mm-sized glass beads were added above the 5 mm glass bead. Just prior to initiation of an experiment, one Azulfidine® tablet was mounted on a holder whereas two L-870,810 capsules or three Asacol® tablets were put on top of the 1 mm-sized glass beads. On top of the cell, a combination of three glass fiber filters (MNGF-5, 0.4 μm pore size, 25 mm diameter; Macherey-Nagel, Germany), followed by 0.1 g glass wool and one more MNGF-5 glass fiber filter was used. Volume of dissolution medium was 50 mL during Stage 1 and 250 mL during Stage 2 (Fig. 2).

In all cases, experimental conditions and sample treatment were as described above for the evaluation of dissolution of Azulfidine® tablets in upper intestinal lumen. However, online filtration when using the flow through apparatus was not adequate for complete separation of solid particles, and additional filtration through 0.45 μm regenerated cellulose filters was necessary.

Analytical Methods

Sulfasalazine concentrations were measured with HPLC-UV. An Agilent Eclipse XDB-C8 column (4.6 × 150 mm², 5 μm) was used. The mobile phase consisted of water:methanol (55:45, v/v). The flow rate was 0.4 mL/min. The detection wavelength was 365 nm. Injection volume was 20 μL. Elution time was approximately 5.5 min and limit of detection was 91 ng/mL. Slopes of calibration curves did not differ significantly from each other by accepting a Type I error of 0.05.

L-870,810 was assayed as described previously.13

Mesalamine concentrations were measured with HPLC-UV. A LiChrospher® 100 RP-18 endcapped, 5 μm, 250 × 4.6 mm² column (Merck KGaA, Darmstadt, Germany) was used. The mobile phase consisted of phosphate buffer pH 7.4: methanol (95:5, v/v). The flow rate was 1.0 mL/min. The detection wavelength was 331 nm. Injection volume was 20 μL. Elution time was approximately 6 min and limit of detection was 182 ng/mL. Slopes of calibration curves did not differ significantly from each other by accepting a Type I error of 0.05.

Data Analysis

Immediate Release Dosage Forms

Evaluation of the dissolution process was performed assuming that the process is controlled by diffusion, the dissolving particles remain spherical with similar and continuously decreasing size during the process, and the diffusion layer thickness, δ, equals the radius, r, of the dissolving particle, by using the following equation13:

()

where X_d is the amount dissolved at time t, V is the volume of the medium, X is the amount remaining to be dissolved, X_s is the amount that saturates the dissolution medium, and z_{δ = r} the dissolution rate constant

()

where D is the diffusion coefficient, N is the number of dissolving particles, Γ is the shape factor, and ρ is density of the dissolving particle. It should be noted that z_{δ = r} has units [mL mg^−1/3 h⁻¹]. In this study, the value of z_{δ = r} was used for evaluating differences in the dissolution process of a specific immediate release dosage form and describing the dissolution process in the model dependent in silico simulations. The value of z_{δ = r} corresponding to a specific individual data set was estimated as described recently.13 If plateau was reached, all data points up to the plateau level were used and plateau level was estimated by averaging relevant data points. If plateau was not reached, all data were used and the plateau observed under other type/intensity of convection conditions in the same medium was used. It is noted that estimation of z_{δ = r} value in the medium simulating the conditions in the ascending colon is not affected by the starting level of dissolution profile.

Data in FeSSCoF could not be collected (Azulfidine®) or were not collected (L-870,810 granules, because of the lack of human data in the fed state). As composition of SIF_ileum does not change with dosing conditions (Tables 1 and 2), only z_{δ = r} values estimated from data collected under conditions simulating the fasted state were analyzed with analysis of variance.

Three factors were considered: the type/intensity of fluid convection (mini-paddle vs. flow-through, Factor A), the level of simulation of luminal composition in the lower intestine in the fasted state (Level I vs. Level II biorelevant media simulating fasted state conditions, Factor B), and regional differences in luminal composition of the lower intestine in the fasted state (distal ileum vs. ascending colon in the fasted state, Factor C).

In the case of Azulfidine®, z_{δ = r} values from data collected with the mini-paddle apparatus in FaSSCoF could not be estimated reliably because of the rapid dissolution process. Thus, z_{δ = r} values were analyzed with two, two-factor analyses of variance. Values estimated from data in SIF_ileum were analyzed for the impact of type/intensity of fluid convection (Factor A), and level of simulation of composition at distal ileum (Factor B). Values estimated from data collected with the flow-through apparatus were analyzed for the impact of level of simulation of composition at a specific region of the lower intestine in the fasted state (Factor B), and regional differences in luminal composition of the lower intestine in the fasted state (Factor C).

In the case of L-870,810 granules, z_{δ = r} values were analyzed with three-factor analyses of variance (Factors A, B, and C).

In all cases, significance was evaluated accepting Type I error of 0.05. Tukey's post hoc test was applied, where needed. All relevant works were performed with Sigmastat (Ver.3.5; Systat Software Inc., San Jose, California).

Sulfasalazine data were qualitatively compared with previously collected sulfasalazine data in the plasma of healthy adults, after single dose administrations in the fasted and in the fed state.24

Data of L-870,810 granules were used for simulating the arrival of the API in plasma based on a recently proposed physiologically based pharmacokinetic model.13 The model accounts for dissolution in the stomach and intestine and potential for precipitation, along with accounting for GI transit and simultaneous absorption. Simulations were performed by using the STELLA® 9.0.2 software (Isee Systems, Inc., Lebanon, New Hampshire). Compared with the previous model,13 the model applied in the present study has the following differences:

• It includes the z_{δ = r} values that were estimated by the dissolution data collected under conditions simulating the environment in distal ileum and in ascending colon,
• It distinguishes intestinal residence into three different locations, upper small intestine (1.5 h), distal ileum (1 h) and ascending colon (3.5 h) instead of two locations assumed in the previous study (1.5 h in the upper intestine and 4.5 h in the lower intestine) and, thus, without affecting the total duration of absorption estimated previously from pharmacokinetic data,13 and
• It takes into account differences in solubility in distal ileum and ascending colon (estimated from the plateau levels of dissolution profiles, this study), and differences in the effective surface area in small intestine and ascending colon [based on differences in luminal fluid volumes (100 mL, small intestine/40 mL, colon) and radii (1.5 cm, small intestine/2 cm, colon)]. No magnification of the cylindrical area of the ascending colon was assumed, as it lacks folds and villi and being a highly permeable API, L-870,810 is considered to be absorbed from the tips of the microvilli.

Predicted plasma levels were evaluated versus individual actual plasma data from a previously conducted clinical study, after administration of 800 mg granules of L-870,810 (400 mg free acid) to healthy fasted adults.13

Asacol®

Dissolved amounts of mesalamine remaining in the lower intestine and cumulative amounts of mesalamine entering the mucosa of the lower intestine were estimated by numerical convolution using in vitro dissolution data collected in this study with Level II biorelevant media. Convolution was performed assuming first order kinetics for mesalamine transport into the colonic mucosa. First-order absorption rate constant, k_a, was estimated using the following equation

()

where P_eff is the effective permeability coefficient, SA is the effective surface area, and V the effective volume of luminal contents. A cylindrical shape of lower intestine was assumed. As mesalamine is a low permeability API, a magnification factor of 6.5 was assumed to take into account the increase of surface area because of the microvilli in the colon,27 and Eq. 3 becomes

()

where r is the colonic radius which was assumed to be 2 cm. Previously published effective permeability coefficients were considered in relevant estimations [P_eff = (0.04 ± 1.36) × 10⁻⁶ cm/s].16 These values have been estimated from data collected with rat ileal perfusions by using concentrations of similar level to what is expected to occur in the colon of humans after three Asacol® tablets (400 mg/tab), that is, greater than 1 mg/mL.28 Because of the high variability of the reported P_eff values, P_eff and P_eff + 2SD values were used in relevant estimations.

All estimations were performed using SigmaPlot v.11.0 software (Systat Software Inc.).

Evaluation of Sulfasalazine Solubility in Intestinal Lumen and of Azulfidine® Tablets Dissolution in Upper Intestinal Lumen

Sulfasalazine solubility data in Level I and II SIF_ileum, FaSSCoF, and FeSSCoF are given in Table 3. Level of simulation of composition in the lower intestine did not affect the data. Although a pH effect is apparent (solubilities in FeSSCoF are lower than in SIF_ileum or FaSSCoF), a solubilization effect in FaSSCoF could also be claimed [solubilities in SIF_ileum (pH 7.5) vs. solubilities in FaSSCoF (pH 7.8)], in line with previous data with other lipophilic APIs in FaSSCoF.6

Estimated z_{δ = r} values for the dissolution of Azulfidine® tablets in Level II FaSSIF-V2 and in Level II FeSSIF-V2 with the mini-paddle apparatus (Fig. 3) were 277 ± 33 mL mg^−1/3 h⁻¹, and 227 ± 145 mL mg^−1/3 h⁻¹, respectively. Based on the plateau levels observed in FaSSIF-V2 and in FeSSIF-V2 (Fig. 3), the ratio dose/solubility is more than 400 mL in these media and, therefore, it can be argued that dissolution of the dose in upper intestine is incomplete, at least in the fasted state.

Two-Stage Single-Compartment Models for Evaluating Dissolution of Azulfidine® Tablets in Distal Ileum and Proximal Colon

In SIF_ileum, the impact of type/intensity of fluid convection and the level of simulation of conditions in distal ileum on z_{δ = r} values (Table 4) was not significant (p ≥ 0.275). Although the power of the test was low, even if the impact was significant it would have no practical importance because of the rapid completion of the process in all cases (Figs. 4 and 5).

In FaSSCoF, dissolution data with the mini-paddle apparatus could not be discussed because of the lack of data points prior to the plateau level (Fig. 4). Data with the flow-through apparatus indicated that the level of simulation of luminal conditions affects z_{δ = r} values (Table 4) significantly (p = 0.002), whereas the difference between data in SIF_ileum and FaSSCoF was also significant (p = 0.003).

Under fed state simulating conditions, the ratio [solubility in SIF_ileum/solubility in FeSSCoF] (Table 3) is 4.6 and 4.0 in Level I and Level II biorelevant media, respectively, that is, slightly smaller than the ratio [volume of FeSSCoF/volume of SIF_ileum] (5/1, Table 2) and no dissolution profile could be obtained in FeSSCoF (Fig. 5).

It is interesting to note that, using the mini-paddle apparatus, estimated mean z_{δ = r} values in Level II SIF_ileum (Table 4) are about four times smaller than in Level II FaSSIF-V2 (69 vs. 277 mL mg^−1/3 h⁻¹) and more than three times smaller than in Level II FeSSIF-V2 (69 vs. 227 mL mg^−1/3 h⁻¹).

Two-Stage Single-Compartment Models for Evaluating Dissolution of L-870,810 Granules in Distal Ileum and Proximal Colon

In line with previous solubility data in human aspirates from the upper GI lumen in the fasted state and dissolution data in media simulating the conditions in upper GI lumen in the fasted state,13 dissolution of L-870,810 granules in SIF_ileum and in FaSSCoF was limited (Fig. 6).

Compared with Azulfidine® tablets, dissolution of L-870,810 granules in SIF_ileum and in FaSSCoF was slower (Fig. 6 vs. Fig. 4). Three-way ANOVA confirmed that z_{δ = r} values for L-870,810 granules (Table 5) are affected significantly by difference in composition between SIF_ileum and FaSSCoF, by the type/intensity of fluid convection, and by their interaction (p < 0.001 for all relevant comparisons). However, z_{δ = r} values of L-870,810 granules are not affected significantly neither by the level of simulation of composition in lower intestine nor by the interaction of level of simulation with another factor. Therefore, differences in dissolution of L-870,810 granules in distal ileum and ascending colon are primarily due to pH differences with the effect of other luminal components being non-significant.

Interestingly, using the mini-paddle apparatus, estimated mean z_{δ = r} values in Level I SIF_ileum (Table 5) are more than five times smaller than in Level I FaSSIF-V2 (121 mL mg^−1/3 h⁻¹)13 and in Level II SIF_ileum (Table 5) are at about 38 times smaller than in Level II FaSSIF-V2 (377 mL mg^−1/3 h⁻¹).13 In contrast, estimated mean z_{δ = r} value in Level I FaSSCoF (Table 5) is 2.4 times bigger than in Level I FaSSIF-V2 (121 mL mg^−1/3 h⁻¹)13 and in Level II FaSSCoF (Table 5) is slightly smaller than in Level II FaSSIF-V2 (377 mL mg^−1/3 h⁻¹).13

Two-Stage Single-Compartment Models for Evaluating Dissolution of Asacol® Tablets in Distal Ileum and Proximal Colon

In line with data collected with the immediate release products with the two apparatus used in the present study, dissolution of Asacol® tablets using the mini-paddle apparatus was not affected by the level of simulation of luminal composition in the fasted state (data not shown). Consequently, experiments with Asacol® were completed with Level II simulated luminal fluids only. In SIF_ileum, amounts dissolved had low variability but they reached solubility level only with the mini-paddle apparatus (Fig. 7). The lower profiles obtained with the flow-through apparatus indicate an impact of type/intensity of SIF_ileum convection on dissolution kinetics. Data in FaSSCoF and FeSSCoF also indicate an impact of type/intensity of fluid convection; in addition, they indicate a negative food effect on dissolution of Asacol® tablets both with the mini-paddle and the flow-through apparatus (Fig. 7), primarily due to pH differences between FaSSCoF and FeSSCoF.