Development of a Solid Supersaturable Micelle of Revaprazan for Improved Dissolution and Oral Bioavailability Using Box-Behnken Design

Abstract

Purpose

To enhance the oral bioavailability of revaprazan (RVP), a novel solid, supersaturable micelle (SSuM) was developed.

Methods

Surfactants and solid carriers were screened based on a solubility and a flowability test, respectively. Supersaturating agents, including Poloxamer 407 (P407), were screened. The SSuM was optimized using a Box-Behnken design with three independent variables, including Gelucire 44/14:Brij L4 (G44/BL4; X₁) and the amounts of Florite PS-10 (FLO; X₂) and Vivapur 105 (VP105; X₃), and three response variables, ie, dissolution efficiency at 30 min (Y₁), dissolution enhancing capacity (Y₂), and Carr’s index (Y₃). The solid state property was evaluated, and a dissolution test was conducted. RVP, Revanex^®, solid micelle (P407-free from the composition of SSuM), and SSuM were orally administrated to rats (RVP 20 mg equivalent/kg) for in vivo pharmacokinetic study.

Results

G44 and BL4 showed great solubility, with a critical micelle concentration range of 119.2–333.0 μg/mL. P407 had an excellent supersaturating effect. FLO and VP105 were selected as solid carriers, with a critical solidifying ratio (g/mL) of 0.30 and 0.91, respectively. With optimized values of X₁ (–0.41), X₂ (0.31), and X₃ (–0.78), RVP (200 mg)-containing SSuM consisting of G44 (253.8 mg), BL4 (106.2 mg), FLO (99.3 mg), VP105 (199.8 mg), and P407 (40 mg) was developed, resulting in Y₁ (40.3%), Y₂ (0.008), and Y₃ (12.3%). RVP existed in an amorphous state in the optimized SSuM, and the SSuM formed a nanosized dispersion in the aqueous phase, with approximately 71.7% dissolution at 2 h. The optimized SSuM improved the relative bioavailability of RVP in rats by approximately 478%, 276%, and 161% compared to raw RVP, Revanex^®, and solid micelle, respectively.

Conclusion

The optimized SSuM has great potential for the development of solidified formulations of poorly water-soluble drugs with improved oral absorption.

Keywords:

• revaprazan
• supersaturation
• solid micelle
• Box-Behnken design
• dissolution
• oral bioavailability

Introduction

Revaprazan (RVP), a new acid-suppressive agent, reversibly inhibits H+/K+-ATPase.¹ Because of its prolonged action as an alternative proton pump inhibitor, RVP has been recently adopted for the treatment of gastroesophageal reflux disease and gastric ulcer.² After absorption in the gastrointestinal (GI) tract, RVP accumulates in the canaliculi of parietal cells.² In the acidic environment around the parietal cells, RVP activates and exerts its pharmacological activity. RVP shows linear pharmacokinetics after oral administration at doses of 2–30 mg/kg and increases the intragastric pH in a dose-dependent manner.³,⁴ RVP, classified as a class II drug according to the biopharmaceutical classification system (BCS), has limited water solubility and therefore low oral bioavailability (BA).⁵,⁶ To improve the solubility and oral BA of RVP, various approaches, such as solid dispersion, microspheres, and self-microemulsifying drug delivery systems, have been used.^7–9

Lipid-based formulations have been spotlighted as a means to enhance the dissolution and oral absorption of poorly water-soluble drugs.¹⁰ Among them, micelles have been recognized as one of the most promising drug delivery systems.¹¹ In the aqueous phase, amphipathic surfactants spontaneously assemble at concentrations above the critical micelle concentration (CMC), forming micellar structures that can accommodate hydrophobic molecules in their cores. Thus, micelles can enhance the dissolution and oral absorption of poorly water-soluble drugs.¹² Further solubilization of poorly soluble drugs can be achieved by incorporating additional strategies, such as supersaturation.¹³

Supersaturating agents render water-insoluble drugs soluble at concentrations above equilibrium solubility, without precipitation.¹⁴ Achieving an increased drug concentration in the supersaturated state in the intestinal lumen would enhance absorption. A number of hydrophilic supersaturating agents, including polyvinylpyrrolidones and poloxamers, have been introduced as precipitation inhibitors.¹⁵ These supersaturating agents have been verified to kinetically and/or thermodynamically prolong the supersaturated state by inhibiting crystal growth or altering the microclimate conditions, where drug molecules are enveloped with an aqueous phase.¹⁶ Theoretically, optimal formulations should maintain the supersaturated state for a sufficient period of time to allow enhanced absorption. Silybin-loaded, supersaturable, polymeric micelle using Soluplus and Copovidone showed 4.5-fold increased oral BA compared to free silybin in rats.¹⁷

Most conventional micellar formulations are in the liquid form and have limitations of drug leakage and capsule aging.¹⁸,¹⁹ Recently, solid carriers have been introduced to generate solidified micelles, in which water-insoluble drug molecules exist in an amorphous state, resulting in a high dissolution rate.²⁰ After reaching the GI tract, solid micelles release the micellar components that form micellar structures due to peristalsis of the GI tract thus enhancing the absorption of poorly water-soluble drugs.²⁰,²¹ When developing a solidified micelle, various variables, such as micelle composition and mass of the solid carrier, have to be optimized. The response surface methodology, which combines mathematical and statistical approaches to suggest optimized response variables, has been used to optimize pharmaceutical formulations.²²,²³ In particular, the Box-Behnken design is a well-known experimental method for formulation optimization because it efficiently suggests an accurate response with fewer combinations.²⁴

In the present study, we aimed to develop a novel RVP-loaded solid supersaturable micelle (SSuM). To this end, RVP-dissolved supersaturable micelle (SuM), composed of surfactant mixtures, was solidified with selected solid carriers. A three-level Box-Behnken design was employed to optimize the composition, using three independent variables, ie, the ratio of surfactants and the amounts of silica-based adsorbent and cellulose-based diluent. Optimized SSuM revealed excellent characteristics in terms of flowability, dissolution, and oral BA in rats. Thus, we expect that this SSuM system would be useful for practical development of a solid dosage form with enhanced dissolution of poorly soluble drugs.

Materials and Methods

Materials

RVP powder (purity >99%) and Revanex^® tablets containing 200 mg RVP were supplied by Yuhan (Seoul, Korea). Aerosil^® 200 was supplied by Evonik Degussa GmbH (Frankfurt am Main, Germany). Avicel PH 101, Brij L4 (BL4), Tween^® 20, and Tween^® 80 were purchased from Sigma-Aldrich (St. Louis, MO, USA). Florite PS-10 (FLO; calcium silicate) was purchased from Tomita Pharmaceutical Co., Ltd. (Tokushima, Japan). Gelucire 44/14 (G44), Labrafil^® M 1944 CS, Labrasol^®, and Lauroglycol^® 90 were gifted by Gattefossé (Saint Priest, France). Kolliphor^® EL, Kolliphor^® RH40, Kolliphor^® RH60, Kollidon K30, Kollidon K90, Poloxamer (P)188, and P407 were obtained from BASF (Ludwigshafen, Germany). Low-substituted hydroxypropyl cellulose B1 (L-HPC) was supplied by Shin-Etsu Chemical Co., Ltd. (Tokyo, Japan). Neusilin^® US2 (magnesium aluminometasilicate) was supplied by Fuji Chemical Industry Company (Toyama, Japan). Sylysia^® 350 (porous silica) was supplied by Fuji Silysia Chemical Co., Ltd. (Aichi, Japan). Vivapur (VP)12 and VP105 were supplied by JRS Pharma (Rosenberg, Germany). High-performance liquid chromatography (HPLC)-grade methanol was purchased from JT Baker (Phillipsburg, NJ, USA). All other chemicals used were of analytical grade.

Animals

Sprague-Dawley rats (Male, 200–250 g, 7–9 weeks) were purchased from Orient Bio (Gyeonggi-do, Korea). Prior to the experiments, all rats were allowed to adapt to our legal animal care facility for one week. All rats were fasted overnight for 12–18 h before drug administration, while they freely accessed to water. All animal trials were carried out in compliance with the National Institutes of Health guidelines on the care and use of laboratory animals and were permitted by the Institutional Animal Care and Use Committee of Chung-Ang University (protocol no. 2020–00062), Seoul, Korea.

Solubility Test

The equilibrium solubility of RVP in various vehicles was measured to screen a suitable surfactant. Briefly, an excess of RVP was added into 1 mL of 1% surfactant solutions. Each solution was shaken on a mechanical shaker (CM-1000; EYELA, Tokyo, Japan) at 25°C for 24 h. After reaching equilibrium, the mixtures were centrifuged (Smart R17; Hanil Science Industrial, Incheon, Korea) at 12,000 ×g for 10 min to exclude undissolved RVP. The supernatant was filtered through a 0.45-μm nylon membrane filter (Whatman International, Kent, UK) and appropriately diluted with methanol. Each sample was analyzed by HPLC to determine the concentration of RVP. To determine the CMC, aqueous solutions of G44, BL4, and a G44/BL4 mixture (50:50 [w/w]) were prepared and serially diluted. Then, the solubility of RVP at each concentration was measured as described above.

HPLC Analysis of RVP

RVP was quantified by HPLC as previously reported.⁶ Twenty microliters of each sample was injected into the HPLC system (e2695; Waters, Milford, MA, USA) consisting of a pump (W2690/5; Waters), an ultraviolet detector (W2489; Waters), and a data station (Empower 3; Waters). An isocratic mobile phase consisting of 93% of methanol in distilled water and a C18 column (150 × 4.6 mm, 5 μm; Agilent Technologies, Santa Clara, CA, USA) was used for chromatographic separation. The flow rate was 1.0 mL/min and the column temperature was set to 35°C, then RVP detection was achieved at 270 nm.

Preparation of RVP-Loaded SuM and SSuM

Based on a previous report,²⁰ drug-free micelles were prepared by mixing G44 and BL4 at 60°C. Various supersaturating agents (10%, w/w) were added to the micelles and homogeneously mixed to prepare SuM, in which RVP (200 mg) was dissolved. To prepare SSuM, RVP-loaded SuM was adsorbed into the powder blend of FLO and VP105, and thoroughly kneaded. Then, the paste was granulated using a high-speed kneading granulator (Triple Master TMGV-1; Shinagawa Machinery Works, Nara, Japan) at a revolution speed of 40 rpm for 20 min, and the granules were passed through a 1000 μm mesh sieve. To avoid unintentional water involvement during granulation process, the obtained fine granules were dried in an oven at 60°C for 12 h (no weight loss was observed).

In vitro Dissolution Study

According to the USP apparatus II (paddle) method, dissolution tests for each sample were performed using a Vision Classic 6TM dissolution tester and a Vision heater. Each formulation containing 200 mg RVP equivalent was dispersed in 900 mL of distilled water at 37 ± 0.5°C and stirred at 50 rpm. Five milliliters of sample were taken at predetermined time points (5, 15, 30, 60, 90, and 120 min) and filtered through a 0.45 μm nylon membrane filter. An equivalent volume of fresh medium was added to maintain a constant dissolution environment. The filtrates were appropriately diluted with mobile phase, and the amount of dissolved RVP was quantified by HPLC as described above. To compare the dissolution behavior between formulations, the dissolution efficiency (DE) and dissolution enhancing capacity (DEC) were calculated using the trapezoidal rule as follows: DE (%) = , where y is the percentage of dissolved product; DEC = (DE_SSuM-DE_RVP)/W_SSuM, where DE_SSuM and DE_RVP are the DE of RVP-loaded SSuM and raw RVP, respectively, and W_SSuM is the amount (mg) of SSuM.²⁵

Screening of Solid Carriers

To obtain SSuM, two types of solid carriers were screened at different weight ratios: Aerosil^® 200, Neusilin^® US2, Sylysia^® 350, and FLO as silica-based adsorbents (0.25–0.5 weight ratio to SuM); VP12, VP105, L-HPC, and Avicel PH 101 as cellulose-based diluents (0.77–2.0 weight ratio to SuM). Solidifying behavior was determined by the levigation method as previously reported.²⁶,²⁷ Briefly, solid carriers were added and mixed with a fixed amount of drug-free SuM (G44/BL4 mixture, 50:50 [w/w]) in a mortar, and blending was discontinued when a non-flowing clot was formed. The physical property of the solidified products was evaluated in terms of flow behavior.

Determination of Flow Behavior

The flow property of the solidified mass was evaluated using Carr’s index (CI), which was calculated using the following equation: CI (%) = [ρ(tapped)-ρ(bulk)]·100/ρ(tapped), where ρ(tapped) and ρ(bulk) refer to the tapped and bulk densities, respectively. Apparent bulk and tapped bulk densities were measured using the cylinder method. Briefly, accurately weighed granules were poured into a cylinder and the volume was read to calculate the apparent bulk density. Another sample was tapped 100 times for tapped bulk density using a powder tester (ABD-100; Tsutsui Scientific Instruments, Tokyo, Japan). The critical solidifying ratio (CSR; g/mL), defined by the minimum required mass (g) of solid carrier to solidify a 1 mL of drug-free SuM, was obtained using a critical constant value of CI.²⁸

Optimization of RVP-Loaded SSuM Using a Box-Behnken Design

To optimize the composition of the SSuM, the Box-Behnken design was employed. Three components were included as independent variables: the ratio of surfactants (G44/BL4; X₁) and the amounts of FLO (X₂) and VP105 (X₃), which were set within ranges of 1:0–0:1, 60–120 mg, and 180–360 mg, respectively. To obtain the optimized formulation with high desirability function, three response variables including DE at 30 min (DE₃₀; Y₁), DEC (Y₂), and CI (Y₃) were adopted. The design consisted of 15 experimental points to find a model fit, and mathematical correlations between inputs and outputs were evaluated using Minitab software (ver. 18.0; Minitab, State College, PA, USA).

Droplet Size Measurement

The droplet size of SSuM dispersed in distilled water was measured using a dynamic light-scattering particle size analyzer (Zetasizer Nano-ZS; Malvern Instruments, Worcestershire, UK). SSuM containing equivalent to 200 mg RVP was accurately weighted and dispersed in 250 mL of distilled water and gently vortexed to obtain a homogeneous dispersion. The mixture was centrifuged at 14,000 ×g for 5 min to completely remove undissolved solid carriers. Then, all samples were transferred to a transparent cuvette, placed in a thermostatic chamber, and light scattering was monitored with a 50-mV laser at an angle of 90° at 25°C.

Solid-State Property Assessment

The solid-state properties of free RVP, FLO, VP105, a physical mixture (PM) of RVP and solid carriers, and optimized SSuM were inquired using scanning electron microscopy (SEM), differential scanning calorimetry (DSC), and powder X-ray diffractometry (PXRD). Morphological features of each sample were visualized using a scanning electron microscope (Sigma 300; Carl Zeiss Meditec AG, Jena, Germany). The samples were put on a brass disc using double-sided adhesive carbon tape. Employing a Hitachi ion sputter (E-1030), each sample was coated with platinum under vacuum for 120 s at a current of 4 mA. The samples were scanned at 5 kV. The thermal characteristics of each powder were evaluated using a DSC-Q20 calorimeter (TA Instruments, New Castle, DE, USA). After sealing in an aluminum pan, each sample (2–3 mg) was subjected to heating in the range of 50°C to 260°C at a rate of 5°C/min under a constant nitrogen flow at 25 mL/min. PXRD patterns of the samples were determined using an X-ray diffractometer (D8 Advance; Bruker, Mannheim, Germany) with nickel-filtered Cu Kα radiation. The diffractogram was acquired by scanning at the 2θ range of 5–60º, with a step angle of 0.02º and scanning speed of 5º/min.

In vivo Oral Absorption Study

Oral Administration and Plasma Sampling

Rats were randomly 4-grouped (n = 6–8 rats per group): group 1 for raw RVP, group 2 for Revanex^® (reference drug), group 3 for solid micelle (P407-free from the composition of SSuM), and group 4 for the optimized SSuM. The tablet form of Revanex^® was ground into powder for oral intake. Of all preparations, an equivalent dose of 20 mg/kg RVP was accurately weighed and filled into hard gelatin capsules (Torpac^® capsule size 9; Torpac, Fairfield, NJ, USA). Each capsules were administered directly into the stomach lumen using a dosing syringe plunger (Torpac^® Kit, Torpac) and then, 1 mL of distilled water was injected orally to encourage the micelle formation. Blood samples of approximately 0.3 mL were collected from the retro-orbital plexus using heparinized capillary tubes at fixed time points (0.33, 0.67, 1, 1.5, 2, 4, 6, 8, 12, and 24 h) and underwent centrifugation at 16,000 ×g for 10 min for plasma separation. The plasma was transferred to a microtube and rapidly frozen at –80°C until liquid chromatography-electrospray ionization-tandem mass spectrometry (LC-ESI-MS/MS) analysis.

Determination of RVP in Plasma Samples by LC-ESI-MS/MS

Fifty microliters of plasma samples were mixed with 1 mL of internal standard solution (50 ng/mL pioglitazone in methanol) and the mixture was vortexed for 3 min. After centrifugation at 16,000 ×g for 5 min, 50% methanol (800 μL) was added to 200 μL of the supernatant. The mixture was vortexed and centrifuged at 16,000 ×g for 5 min. Finally, supernatant (100 μL) was transferred into a vial for introduction into the LC-ESI-MS/MS system. LC separations were achieved using a Vanquish Duo UPLC (Thermo Scientific, MA, USA). One microliter of each sample was separated using a Hypersil GOLD dC18 column (100 × 2.1 mm, 1.9 μm; Thermo Scientific) at 40°C. An isocratic mobile phase consisting of distilled water, methanol, and formic acid (30:70:0.05, [v/v/v]) was used at a flow rate of 0.2 mL/min. Eluted components were delivered into an Orbitrap Q Exactive Plus mass spectrometer (Q-EXACTIVE PLUS orbitrap; Thermo Scientific) equipped with a heated ionization source operated in positive ion mode. The operating parameters were as follows: electrospray voltage, 3.5 kV; sheath gas flow rate, 35 (arbitrary units); auxiliary gas, 10 (arbitrary units); sweep gas, 0 (arbitrary units); auxiliary gas heater temperature, 350°C; and capillary temperature, 350°C.

Pharmacokinetic (PK) Assessment

PK parameters including area under the curve from 0 to 24 h (AUC_0–24h), maximum plasma concentration (C_max), and time to reach the C_max (T_max) were analyzed using the BA Calc 2007 PK analysis program (Ministry of Food and Drug Safety, Chungcheongbuk-do, Korea). AUC_0–24h was calculated by the linear trapezoidal rule and both C_max and T_max were calculated directly from the plasma concentration-time data. To compare oral absorption between formulations, relative BA was computed by dividing the AUC_0–24h of the samples by that of the reference.

Statistical Analysis

All data are represented as the mean ± standard deviation (SD). Statistical significance was evaluated using Student’s t-test, and p <0.05 was considered statistically significant.

Results and Discussion

Screening of Solubilizing Vehicles

The selection of a suitable solubilizing vehicle is critical for micelle formation, as RVP has poor water solubility (approximately 230 μg/mL in distilled water). The solubility of RVP in various surfactant solutions is shown in Figure 1A. Among the surfactants tested, G44 showed the greatest solubility (1,489 μg/mL) followed by BL4 (1,342 μg/mL). G44 alone did not form a homogenous dispersion (polydispersity index (PDI) >0.6, data not shown); however, the addition of BL4 resulted in a homogeneous and nano-sized dispersion. This might be attributed to the difference in the hydrophilic–lipophilic balance value between G44 (14) and BL4 (9). As BL4 is more hydrophobic than G44, it could efficiently enclose RVP. Both surfactants are biocompatible and have been widely used in nanoemulsion formulations.²⁹,³⁰ Therefore, a mixture of G44 and BL4 was selected as a vehicle for micellization.

Figure 1 Solubility of RVP in various surfactant solutions (1% [w/w]) (A) and CMC determination via the solubilization method (B).

Abbreviations: LG, Lauroglycol 90; LF, Labrafil M 1944 CS; BL4, Brij L4; LS, Labrasol; RH60, Kolliphor RH60; RH40, Kolliphor RH40; G44, Gelucire 44/14; EL, Kolliphor EL; T80, Tween 80; T20, Tween 20.

To determine the amount of mixed surfactants required for optimal micelle formation, the CMC of the mixture was measured by the solubilization method.³¹ As depicted in Figure 1B, at low concentration (~100 μg/mL), the surfactants did not increase the solubility of RVP; however, the solubility sharply increased at high surfactant concentrations (>300 μg/mL), indicating the formation of surfactant micelles. The CMC values were determined as the intersection of the regression lines obtained from the two linear portions of the plot:³² 119.2 μg/mL for G44 alone, 333.0 μg/mL for BL4 alone, and 208.5 μg/mL for the G44/BL4 mixture. Based on these values, to guarantee a sufficiently high concentration above the CMC, G44/BL4 was used at 400 μg/mL in the dissolution experiment.

Screening of Supersaturating Agents

To further enhance RVP solubilization, supersaturating agents were screened. Based on previous reports, the amount of polymers was set to 10% [w/w] of the total amount of vehicle.²⁵,³³ Hydrophilic polymers can induce supersaturation and maintain the supersaturated state by inhibiting nucleation and crystallization via steric stabilization and/or specific interactions, such as hydrogen bonding, hydrophobic interactions, and complex formation.¹⁵,³⁴ Poloxamers and polyvinylpyrrolidones have been used as effective supersaturating agents in various formulations, including lipid-based systems.²⁵,³⁵ Dissolution profiles of G44/BL4 micelles in the presence or absence of various supersaturating agents are shown in Figure 2. All formulations tested showed rapid dissolution during the first 30 min and plateaued thereafter. Compared to micelle alone, the addition of supersaturating polymers somewhat increased RVP dissolution, in the order P407 >K90 ≥K188 ≥K30. In particular, the dissolution level of P407-added micelle (approximately 70% for 2 h) was significantly higher than those of other polymer-added micelles and polymer-free micelle. P407 consists of hydrophobic block (polypropylene glycol) and hydrophilic block (polyethylene glycol). Because of its amphiphilic nature, P407 is able to act as a surfactant and squeeze in the colloidal interface.³⁶ Hydrophobic moieties interact with hydrophobic tails of drug-loaded colloid, rigidifying hydrocarbon layer of the interface, resulting in the action of supersaturation.³⁷ Meanwhile, hydrophilic moieties shield the surface of the micelle, generating hydrophilic barrier which hamper the aggregation and/or destruction of the colloid.³⁷ P407 also spontaneously forms micelle and solubilizes water-insoluble drugs, thereby contributing to supersaturation.³⁸ Thus, P407 was selected as a supersaturating agent for SuM formulation.

Figure 2 Dissolution profiles of micelle with or without various supersaturating agents.

Note: *p <0.05 versus Micelle only.

Abbreviations: P407, Poloxamer 407; K90, Kollidon K90; P188, Poloxamer 188; K30, Kollidon K30.

Selection of Solid Carriers

To develop a solid micelle, it is important to select appropriate solid carriers. We screened two types of solid carriers: silica-based adsorbents (Aerosil^® 200, Sylysia^® 350, Neusilin^® US2, and FLO) and cellulose-based diluents (VP105, VP12, L-HPC, and Avicel PH 101). Silica-based adsorbents have a small particle size and high surface area, a porous structure, and high oil-absorption capacity, thereby minimizing the amount required for solidification.³⁹ However, they sometimes entail incomplete desorption of micelle components because of strong interaction between the drug and silicates.⁴⁰ In comparison, cellulose-based diluents are hydrophilic and completely dissolve in aqueous phase, avoiding the problems of incomplete desorption.²⁸ In this regard, a combination of these solid carriers would be preferred. The flow properties of the solidified granules were investigated based on plots of the CI against the ratio of solid carrier to SuM (Figure 3). Silica-based adsorbents were superior to cellulose-based diluents, showing smaller CSR values (range, 0.30–0.44) than cellulose-based diluents (range, 0.91–1.47). As FLO and VP105 showed the lowest CSR values among the silica-based adsorbents and cellulose-based diluents, respectively, they were selected as solid carriers for SSuM formulation.

Figure 3 Plots of the CI against the ratio of solid carrier to SuM to evaluate flow property changes in SuM formulations with silica-based adsorbents (A) and cellulose-based diluents (B).

Note: Arrows indicate the critical solidifying ratio.

Abbreviations: FLO, Florite PS-10; VP105, Vivapur 105; VP12, Vivapur 12; L-HPC, low-substituted hydroxypropyl cellulose B1.

Statistical Analysis Using Box-Behnken Design

To optimize the RVP-loaded SSuM formulation, a three-level Box-Behnken design was employed. In Box-Behnken design, optimal responses are suggested by calculating the effect of each independent factor and their interactions.²⁴ Based on a preliminary study, three factors were selected as independent variables: the ratio of G44 to BL4 (X₁), the amount of FLO (X₂), and the amount of VP105 (X₃). As shown in Table 1, X₁ was set to range from 1:0 to 0:1. The upper limits (1) of X₂ and X₃ were based on their CSR values (120 mg and 360 mg, respectively) and the lower limits (–1) were set to halves thereof. To find an optimal SSuM formulation, three response variables, ie, DE₃₀ (%; Y₁), DEC (Y₂), and CI (%; Y₃) were employed, with Y₁ ranging from 21.23 to 45.56, Y₂ from 0.050 to 0.097, and Y₃ from 10.2 to 20.6 (Table 2). To estimate the effects of single terms and/or interactions between factors, each response was fitted to a linear and polynomial regression (up to quadratic) models, and statistical parameters of each model were evaluated (Table 3). The standard error of regression (S) represents the quantitative average distance between the experimental values and regression curve and provides information on whether the experimental values were adequately fitted to the adopted model.⁴¹ The model exhibiting the lowest S value is recommended. The prediction accuracy of the model was inferred from the R² value, where the model with the highest R² value is suggested. A good model should turn out to be statistically significant (p <0.05). On the other hand, a lack of fit measure should be statistically non-significant, because the measure indicates large deviation of the data from the fitted model.⁴² Based on the statistical parameter results, quadratic models were selected for further optimization.

Table 1 Independent and Response Variables Used in Box-Behnken Design

Independent Variables		Levels
		–1	0	1
X1	G44/BL4	1:0	1:1	0:1
X2	FLO (mg)	60	90	120
X3	VP105 (mg)	180	240	360
Response variables		Goal
Y1	DE30 (%)	Maximized
Y2	DEC	Maximized
Y3	CI (%)	Minimized

Abbreviations: G44, Gelucire 44/14; BL4, Brij L4; FLO, Florite PS-10; VP105, Vivapur 105; DE₃₀, dissolution efficiency at 30 min; DEC, dissolution enhancing capacity DE₃₀, dissolution efficiency at 30 min; DEC, dissolution enhancing capacity; CI, Carr’s index.

Table 2 Combinations of Independent Variable and Experimental Responses from Runs

Run	X1	X2	X3	Y1	Y2	Y3
1	–1	–1	0	38.14 ± 2.15	0.093 ± 0.002	17.9 ± 0.4
2	1	–1	0	28.75 ± 2.78	0.084 ± 0.006	19.6 ± 0.5
3	–1	1	0	26.45 ± 2.76	0.085 ± 0.004	11.5 ± 0.8
4	1	1	0	21.23 ± 4.97	0.062 ± 0.006	12.6 ± 0.4
5	–1	0	–1	42.09 ± 2.13	0.086 ± 0.006	15.4 ± 0.3
6	1	0	–1	33.61 ± 3.58	0.058 ± 0.008	15.8 ± 0.6
7	–1	0	1	34.44 ± 4.16	0.060 ± 0.005	12.4 ± 0.1
8	1	0	1	29.97 ± 7.43	0.050 ± 0.006	13.7 ± 0.5
9	0	–1	–1	45.56 ± 1.87	0.097 ± 0.002	20.6 ± 0.1
10	0	1	–1	31.67 ± 1.72	0.080 ± 0.002	10.5 ± 0.6
11	0	–1	1	33.85 ± 1.28	0.086 ± 0.005	14.9 ± 0.3
12	0	1	1	27.67 ± 2.86	0.063 ± 0.004	10.2 ± 0.4
13	0	0	0	34.60 ± 2.38	0.089 ± 0.005	13.0 ± 0.5
14	0	0	0	36.62 ± 1.05	0.091 ± 0.006	13.4 ± 0.5
15	0	0	0	35.84 ± 0.97	0.091 ± 0.005	12.9 ± 0.9

Notes: X_1, G44:BL4; X₂, FLO; X₃, VP105; Y₁, DE₃₀; Y₂, DEC; Y_3, Carr’s index; Values are presented as the mean ± SD (n = 3).

Table 3 Summary of the Results of Statistical Analysis and Model Equations for the Measured Responses

Models	S	R^2	R^2 (Adj)	p-value	Lack of Fit p-value	Remark
DE30 (%)
Linear	4.333	59.61	56.66	<0.001	<0.001	-
Linear+square	3.243	79.03	75.72	<0.001	0.286	-
Linear+interaction	4.293	63.27	57.47	<0.001	<0.001	-
Quadratic	3.071	82.69	78.4	<0.001	0.835	Suggested
DEC
Linear	0.011	48.61	44.85	<0.001	<0.001	-
Linear+square	0.006	88.60	86.80	<0.001	0.008	-
Linear+interaction	0.011	52.56	45.07	<0.001	<0.001	-
Quadratic	0.005	92.55	90.63	<0.001	0.439	Suggested
CI (%)
Linear	1.217	85.26	84.18	<0.001	<0.001	-
Linear+square	0.925	92.10	90.86	<0.001	<0.001	-
Linear+interaction	0.996	90.85	89.41	<0.001	<0.001	-
Quadratic	0.521	97.69	97.10	<0.001	0.157	Suggested

Abbreviations: DE₃₀, dissolution efficiency at 30 min; DEC, dissolution enhancing capacity; CI, Carr’s index; S, standard error of regression.

Effects of Independent Variables on the Responses in Experimental Design

A normal distribution is considered a crucial assumption for statistical analysis. Normal probability plots for standardized residuals were linear, indicating that the experimental run data were normally distributed (Supplementary Figure S1A).4,⁴³ 4 Pareto charts indicating the relative importance of the individual and interaction effects based on Student’s t-tests are shown in Supplementary Figure S1B. Statistically significant independent factors were extended the dotted bar which positioned at α = 0.05.⁴⁴ The results of the overall analysis of variances for the three response variables with the sum of squares are presented in Table 4. The amount of variability accounted for by the quadratic model response was statistically significant (p <0.05) for all of the response variables. A lack of fit also signified a small deviation of the data from the fitted model, indicating the predicted values were close to observed values of the response variable. The result of the analysis of variance for each predictor is further provided in Supplementary Table S1. Overall, all linear effect parameters (X₁, X₂, and X₃) had a significant impact on each response. Moreover, some polynomial effect parameters were found to be statistically significant as well. Considering linear and polynomial effect parameters, quadratic model was selected for three responses and their polynomial regression equations calculated as follows:

(1)

(2)

(3)

Table 4 Analysis of Variance of the Experimental Responses for Quadratic Model

Source	DF	Y1 (DE30)			Y2 (DEC)			Y3 (CI)
		SS	F	p-value	SS	F	p-value	SS	F	p-value
Model	9	1576.29	18.57	< 0.001	0.009534	48.30	< 0.001	402.43	164.55	< 0.001
Residual	35	330.03	-	-	0.000768	-	–	9.51	-	–
Total	44	1906.32	-	-	0.010302	-	-	411.97	-	-
Lack of fit	3	8.63	0.29	0.835	0.000061	0.93	0.439	1.41	1.86	0.157

Abbreviations: DF, degrees of freedom; SS, sum of squares, DE₃₀, dissolution efficiency at 30 min; DEC, dissolution enhancing capacity; CI, Carr’s index.

Each linear or polynomial regression coefficient reflects not only the magnitude of its effect on the response variables but also the proportional relationship between the input and responses. Positive coefficients indicate positive relationships to each response, whereas negative coefficients indicate negative relationships between the factors and each response.⁴⁵ As shown in Figure 4, both X₂ and X₃ had negative impacts on Y₁ and Y₂. Once liquid formulations underwent solidification, the dissolution rate decreased, as micelle components tend to be incompletely desorbed from the pores of solid carriers.⁴⁶ Especially, both Y₁ and Y₂ changed drastically when X₂ shifted from the lower limit to the upper limit compared to X₃, because of the insolubility of FLO. Two different solid carriers were employed to obtain SSuM. The silica-based adsorbent, FLO, has a large surface area and high oil-absorption capacity because of its mesoporous structure.⁴⁷ Water-insoluble silicates exert disadvantage on the dissolution rate due to molecular adhesion (eg, hydrophobic interaction) between the silanol group and micelle components, which resulted in a negative association with both Y₁ and Y₂.⁴⁰ In contrast, VP105 is completely soluble in water and enables complete desorption of the micelle components into the aqueous medium.²⁸ Nevertheless, X₃ was negatively associated with Y₁ because Y₁ reflected dissolution behavior in the early phase only. Combinations with a high VP105 content showed a gradual increase in dissolution throughout the period, showing a similar dissolution rate as P407-added SuM at 2 h. Moreover, X₃ had a negative impact on Y₂ because of the amount of VP105 added. To obtain fine granules, it would be necessary to add a large quantity of solid carriers, thus increasing the total system quantity. The large amount of solid carriers entailed a large unit quantity, but this would be disadvantageous to DEC, leading to a negative association between Y₂ and X₃. The addition of both solid carriers had a positive effect on granulation, whereby CI values steeply decreased when the amount of solid carriers increased. As a result, negative effects on Y₃ were found for both X₂ and X₃.

Figure 4 Effects of independent factors on response variables: three-dimensional response surface plots of Y₁ (A), Y₂ (B), and Y₃ (C).

The ratio of surfactants also had a statistically significant impact on each response. G44 showed a higher hydrophilic–lipophilic balance value and better RVP solubility than BL4.⁴⁸,⁴⁹ Therefore, the dissolution rate increased with an increasing amount of G44 and hence, X₁ showed a negative correlation with Y₁. Because the unit quantity was not changed by X₁, X₁ was negatively associated with Y₂ as well (Figure 4C). In addition, because G44 is semi-solid at room temperature, solidification is expected to become easier when the proportion of G44 increases.⁵⁰ This explains the positive correlation between Y₃ and X₁.

Optimization of RVP-Loaded SSuM Using a Desirability Function

The three responses were optimized using appropriate targets. Micelle components should be released from SSuM as soon as it reaches the GI tract. A high dissolution rate in the early phase contributes to the rapid absorption of poorly soluble drugs. At this point, DE₃₀, reflecting the early dissolution behavior, was set to be maximized. Y₂ represents the ratio of dissolution to total mass, and a product exhibiting high dissolution with low nit quantity is desirable; thus, Y₂ was set to be maximized. And Y₃ (CI) was set to be minimized to acquire free-flowing granules.

According to the goals of each response, three independent variables were optimized using a desirability function. As shown in Table 5, X₁, X₂, and X₃ were optimized as –0.41, 0.31, and –0.78, respectively, and the desirability function was 0.748, resulting in a composition of 253.8 mg G44, 106.2 mg BL4, 99.3 mg FLO, 199.8 mg VP105, 40 mg P407, and 200 mg RVP. The granules obtained were easily reconstituted in distilled water, resulting in size and PDI values of 181.3 ± 5.8 nm and 0.218 ± 0.027, respectively, indicating a homogeneous, nanosized micelle formation. To evaluate how well the experiment was designed, prediction percentage errors were calculated: the values were lower than 5% for all responses, indicating that optimized SSuM was successfully prepared using the Box-Behnken design.

Table 5 Predicted Values and Experimental Values for the Optimized RVP-Loaded SSuM Formulation

Y Variables	DE30	DEC	CI
Predicted value	38.6	0.09	12.9
Experimental value	40.3	0.088	12.3
Prediction error (%)	4.40	−2.22	–4.65

Notes: Optimized SSuM contains 200 mg RVP, 253.8 mg G44, 106.2 mg BL4, 99.3 mg FLO, 199.8 mg VP105, and 40 mg P407.

Abbreviations: DE₃₀, dissolution efficiency at 30 min; DEC, dissolution enhancing capacity; CI, Carr’s index.

Solid-State Property Evaluation

The solid-state properties of RVP, FLO, VP105, PM, and SSuM were evaluated by SEM, DSC, and PXRD. The morphological aspects of each sample were visualized using SEM (Figure 5A). Raw RVP had a small, irregular, and asymmetrical shape, as previously reported.⁹ Solid carriers appeared to have indefiniteness of their shape and size. In PM, typical morphologies of both solid carriers and raw RVP were present. However, no distinct crystallinity of RVP was observed in SSuM that existed in a lumped mass, indicating that RVP was completely solubilized and absorbed onto the solid carriers. Further, DSC and PXRD analyses were conducted to evaluate the crystalline property of RVP. DSC thermograms are shown in Figure 5B. RVP showed a sharp endothermic peak at near 220°C, which corresponded to its melting point, and was crystalline in nature (205–208°C).⁸,⁹ There were no specific endothermic peaks for both solid carriers, FLO and VP105, as reported.²⁸ PM exhibited a small endothermic peak near the melting point of RVP; however, SSuM yielded no endothermic peak at the melting point of RVP. In PXRD observation (Figure 5C), intrinsic sharp peaks of the free drug were found at several diffraction angles, indicating the crystallinity of RVP.⁸ The solid carriers produced characteristic intrinsic peaks: 29º for FLO and 21º for VP105.⁵¹ PM yielded similar major peaks of RVP, demonstrating that RVP maintained its crystallinity in PM. In contrast, no intrinsic peaks were found for SSuM. This result is consistent with an earlier report that solubilized drugs within a micelle structure exist in an amorphous state, even though the micelle is solidified.²⁰

Figure 5 Solid-state properties of powders. (A) Scanning electron microscopy images. (B) Differential scanning calorimetry thermograms. (C) Powder X-ray diffractometry patterns.

Note: Scale bar = 10 μm.

Abbreviations: RVP, revaprazan; FLO, Florite PS-10; VP105, Vivapur 105; PM, physical mixture of RVP and solid carriers; SSuM, solid supersaturable micelle.

In vitro Dissolution Study

Figure 6 shows the dissolution profiles of raw RVP, Revanex^® as a reference product, RVP-loaded solid micelle, and RVP-loaded SSuM. As RVP is insoluble in the aqueous phase, raw RVP scarcely dissolved in the medium, exhibiting an extremely low dissolution rate throughout the test period. Revanex^® was ground into powder for direct comparison with SSuM granules. Revanex^® powder showed improved dissolution (28.0% at 2 h) compared to raw RVP; however, its dissolution rate was not sufficiently high. In contrast, solid micelle and SSuM elevated the dissolution level, showing approximately 48.7% and 71.7% dissolution at 2 h, respectively. These values were quite consistent with the dissolution results of G44/BL4 micelle and SuM, suggesting that micelle components were completely released from the solid carriers and solubilized the poorly soluble drug. The dissolution level of SSuM was significantly higher (p <0.05) than those of other samples, showing >50% dissolution within 30 min, after which it gradually increased. The SSuM formulation successfully increased the dissolution rate of RVP via spontaneous generation of a micelle structure as well as the supersaturating action of P407. Together, these results suggested that SSuM provides a promising strategy to improve the oral BA of drugs with poor aqueous solubility.

Figure 6 Dissolution profiles of raw RVP, Revaprazan^® (powder), solid micelle, and the optimized SSuM in distilled water.

Note: Values are the mean ± standard error (n = 3).

Abbreviations: RVP; revaprazan; SSuM, solid supersaturable micelle.

In vivo PK Behavior

To evaluate the effect of the optimized SSuM on oral absorption behavior, a PK study was conducted in rats orally administered raw RVP, Revanex^® powder, solid micelle, and the optimized SSuM. The plasma concentration of RVP was plotted against time (Figure 7). At an equivalent dose of 20 mg/kg of RVP, the plasma concentrations of all four formulations gradually increased for 6 h. However, both solid micelle and optimized SSuM showed greater absorption behavior than Revanex^® and raw RVP. Enhanced intestinal permeability to the nano-sized dispersion may have contributed to the increased plasma RVP concentration.⁵² In particular, the absorption pattern of both micellar formulations was biphasic: a sharp increase within 1 h, then gradual absorption up to 6 h. This behavior was consistent with the results of the dissolution study, suggesting that the micelle components rapidly solubilized RVP in the GI tract and continuously provided a favorable condition for absorption. This may be attributed to the absorption-enhancing effect of the surfactants used, BL4 and G44, which have been widely used to enhance the oral BA of hydrophobic materials.²⁹,³⁰ Specifically, throughout the test period, the PK profile of the optimized SSuM was superior to that of the solid micelle, indicating the usefulness of P407 to achieve supersaturation. P407 has been shown to enhance the oral absorption of water-insoluble drugs in numerous lipid formulations.³³,⁵³ Further, because of its amphipathic nature, many pharmaceutical formulations have exploited P407 as an intestinal permeability enhancer.⁵⁴,⁵⁵

Figure 7 Plasma RVP concentration profiles after oral administrations of different formulations to Sprague-Dawley rats at an equivalent dose of 20 mg/kg of RVP.

Note: Values are the mean ± standard deviation (n = 6–8).

Abbreviations: RVP, revaprazan; SSuM, solid supersaturable micelle.

The PK parameters of each sample are listed in Table 6. Compared with raw RVP and Revanex^®, the solid micelle and optimized SSuM formulations showed greater C_max and AUC_0–24h values, which were in the order SSuM > solid micelle > Revanex^® > raw RVP. Based on the AUC_0–24h values, the solid micelle improved the oral BA of RVP by approximately 3.0- and 1.7-fold versus raw RVP and Revanex^®, respectively. This result was comparable with an earlier report that orally administered sulpiride-loaded micelle composed of Tween 80 and propylene glycol increased the intestinal permeability in rats, showing approximately 4-fold greater permeability coefficient compared to that of sulpiride solution.⁵⁶ Furthermore, relative BA of the optimized SSuM was approximately 4.8-, 2.8-, and 1.6-fold greater than those of raw RVP, Revanex^®, and solid micelle, respectively, indicating the importance of supersaturation in addition to micelle formation in the GI tract. Lipid-based formulations containing the supersaturating agent were superior in drug absorption to conventional formulation. For instance, oral BA of indirubin-loaded self-microemulsifying drug delivery system was increased 1.3-fold by the addition of 0.5% PVP K17 as a supersaturating agent.¹⁴ For BSC class II drugs, including RVP, dissolution is a rate-limiting step in drug absorption.⁵⁷ The improved dissolution of RVP in the micelle formulations enabled enhanced oral absorption. In addition, middle chain fatty acid enhances the permeability of particles across the cell layers.⁵⁸ Both surfactants, G44 and BL4, contain lauric acid moieties that enhance intestinal permeability.⁵⁹ Meanwhile, the T_max of SSuM was slightly delayed, indicating that continuous intestinal absorption of RVP occurred via supersaturated micelle formation. Therefore, we suggest that SSuM would be a promising candidate for the clinical development of the BCS II drug such as RVP. However, as the present study designed in a rodent model only, further investigations on allometric scaling and/or human PK evaluation are necessary for clinical application.

Table 6 PK Parameters of RVP Formulations in Rats

Parameters	Raw RVP	Revanex^®	Solid Micelle	SSuM
AUC0–24h (ng·h/mL)	1662.36 ± 547.06	2877.84 ± 564.18*	4940.90 ± 1591.37*,^#	7941.48 ± 1608.03*,^#,†
Cmax (ng/mL)	195.50 ± 82.23	339.50 ± 51.10*	619.35 ± 160.54*,^#	987.54 ± 171.38*,^#,†
Tmax (h)	5.00 ± 2.10	4.92 ± 2.25	4.92 ± 1.86	6.00 ± 1.79
Relative BA (%)
versus raw RVP	–	173.1	297.2	477.7
versus Revanex^®	57.8	–	171.7	276.0
versus Solid micelle	33.6	58.2	–	160.7

Notes: Significantly different at p <0.05, *versus RVP; ^#versus Revanex^®; ^†versus solid micelle.

Abbreviations: AUC, area under the curve; BA, bioavailability; RVP, revaprazan; SSuM, solid supersaturable micelle.

Conclusion

We successfully developed a novel RVP-containing SSuM using a Box-Behnken design with three independent variables, X₁ (G44/BL4), X₂ (FLO), and X₃ (VP105), resulting in DE₃₀ (40.3%), DEC (0.088), and CI (12.3%). With a high desirability function value (0.748), the optimized SSuM reflected a negligible percentage prediction error (<5%). No distinct crystalline of RVP was found indicating that RVP existed in an amorphous state in the optimized SSuM and showed substantially improved dissolution behavior compared to raw RVP. PK study in rats revealed that the SSuM improved relative BA by approximately 4.8- and 2.8-fold versus raw RVP and Revanex^®, respectively. Thus, SSuM shows promising potential for the development of a solid formulation of the poorly water-soluble RVP.

Author Contributions

All authors made a significant contribution to the work reported, whether that is in the conception, study design, execution, acquisition of data, analysis and interpretation, or in all these areas; took part in drafting, revising or critically reviewing the article; gave final approval of the version to be published; have agreed on the journal to which the article has been submitted; and agree to be accountable for all aspects of the work.

Acknowledgments

This work was supported by a grant from the National Research Foundation of Korea (NRF) funded by the Korea government (MSIP) (No. 2019R1A2C2002510). The work was also supported by the Brain Korea (BK) 21 FOUR through the NRF funded by the Ministry of Education of Korea. We also thank to Editage (www.editage.co.kr) for English language editing (CAUNE_6678).

Disclosure

The authors report no conflicts of interest in this work.