Formulation and characterization of self-nanoemulsifying drug delivery systems containing monoacyl phosphatidylcholine

International Journal of Pharmaceutics 502 (2016) 151–160

Contents lists available at ScienceDirect

International Journal of Pharmaceutics journal homepage: www.elsevier.com/locate/ijpharm

Pharmaceutical nanotechnology

Formulation and characterization of self-nanoemulsifying drug delivery systems containing monoacyl phosphatidylcholine Thuy Trana , Xi Xia , Thomas Radesa , Anette Müllertza,b,* a b

Department of Pharmacy, Faculty of Health and Medical Sciences, University of Copenhagen, Universitetsparken 2, 2100 Copenhagen, Denmark Bioneer: FARMA, Faculty of Health and Medical Sciences, University of Copenhagen, Universitetsparken 2, 2100 Copenhagen, Denmark

A R T I C L E I N F O

A B S T R A C T

Article history: Received 22 December 2015 Received in revised form 14 February 2016 Accepted 15 February 2016 Available online 22 February 2016

The study investigated the use of monoacyl phosphatidylcholine (MAPC) in self-nanoemulsifying drug delivery system (SNEDDS). A D-optimal design was used to generate two sets of formulations containing long-chain (LC) or medium-chain (MC) glycerides, caprylocaproyl macrogol-8 glycerides (Labrasol), Lipoid S LPC 80 (LPC) (80% MAPC) and ethanol. The formulations were characterized using dynamic light scattering, microscopy, in vitro lipolysis and viscometric measurements. All LC formulations within the investigated range were predicted to generate polydisperse emulsions while MC formulations generated nanoemulsions with droplet sizes from 23 to 167 nm. Using LPC in MC formulations reduced the nanoemulsion droplet sizes in simulated gastric and intestinal media. The nanoemulsion droplet size of MC SNEDDS containing LPC was not affected by gastrointestinal pH, while the zeta potentials increased at low pH. During in vitro lipolysis, less fatty acids were released when LPC was incorporated into the formulations (2.05  0.02 mmol reduced to 1.76  0.05 mmol when incorporating 30% LPC). Replacing Labrasol by LPC increased the formulation dynamic viscosity from 57  1 mPa s (0% LPC) to 436  8 mPa s (35% LPC) at 25  C, however, this did not considerably prolong the formulation dispersion time. In conclusion, MC SNEDDS containing LPC are promising formulations when desiring to reduce the amount of synthetic surfactants and possibly modify the digestion rate. ã 2016 Elsevier B.V. All rights reserved.

Keywords: Monoacyl phosphatidylcholine Self-nanoemulsifying drug delivery systems Experimental design Dynamic light scattering Cryogenic transmission electron microscopy In vitro lipolysis

1. Introduction The majority of new low-molecular-weight chemical entities in the drug discovery programs of the pharmaceutical industry exhibit insufficient aqueous solubility to ensure adequate oral absorption and therefore often provide poor and unpredictable oral bioavailability (Lipinski, 2000; Lipinski et al., 2012). For several decades, lipid-based drug delivery systems have been developed to overcome this problem by presenting lipophilic drugs in solubilized form in the gastro-intestinal tract, in order to avoid the dissolution process, which is often slow or incomplete with a conventional solid dosage form (Porter et al., 2007). Some other

Abbreviations: Cryo-TEM, cryogenic transmission electron microscopy; DoE, design of experiments; FaSSGFTDC, fasted-state simulated gastric fluid containing sodium taurodeoxycholate; FaMIF, fasted-state median intestinal fluid; LC, longchain; LPC, Lipoid S LPC 80; MAPC, monoacyl phosphatidylcholine; MC, mediumchain; PC, phosphatidylcholine; SNEDDS, self-nanoemulsifying drug delivery systems. * Corresponding author at: Department of Pharmacy, Faculty of Health and Medical Sciences, University of Copenhagen, Universitetsparken 2, 2100 Copenhagen, Denmark. E-mail address: [email protected] (A. Müllertz). http://dx.doi.org/10.1016/j.ijpharm.2016.02.026 0378-5173/ ã 2016 Elsevier B.V. All rights reserved.

bioavailability enhancing mechanisms provided by lipid-based drug delivery systems include prolonging gastric residence time (Van Citters and Lin, 1999), stimulating lymphatic transport (Caliph et al., 2000), increasing intestinal permeability (Buyukozturk et al., 2010; Gupta et al., 2011), reducing intestinal metabolism (Patel and Brocks, 2009) and inhibiting P-glycoprotein activity (Constantinides and Wasan, 2007). Despite these acknowledged advantages, the emulsification of lipid-based formulations often depends on the variable human digestive environments, rendering the dissolution and absorption profile of drugs unpredictable and poorly reproducible (Porter et al., 2007). Using self-nanoemulsifying drug delivery systems (SNEDDS) to form nanoemulsions in gastrointestinal media can overcome these inter- and intra-subject variations (Pouton, 2000). Previous research has shown that SNEDDS reduce the variability in oral absorption and provide higher oral bioavailability for cyclosporine A (Kovarik et al., 1994), probucol (Nielsen et al., 2008), and cinnarizine (Larsen et al., 2013) compared to lipid-based drug delivery systems generating coarse emulsions upon dispersion. Most SNEDDS are comprised of triglycerides, hydrophilic surfactants, co-surfactants and cosolvents, allowing self-emulsification and enhancing drug solubilization (Pouton, 2000). To date,

152

T. Tran et al. / International Journal of Pharmaceutics 502 (2016) 151–160

most surfactants used in SNEDDS are synthetic and potentially provoke irritancy to cell membranes of the digestion tract, especially in children (Gloxhuber, 1974; Attwood and Florence, 1983). Replacing or reducing these synthetic surfactants by a natural surfactant is an interesting option for future development of SNEDDS, particularly as a pediatric dosage form. Monoacyl phosphatidylcholine (MAPC) (Fig. 1), also known as lyso-phosphatidylcholine, is generated from the digestion of phosphatidylcholine in the upper small intestine and absorbed intact into the enterocytes by passive diffusion (Carey et al., 1983; Tso, 1985). As a soluble amphiphilic lipid (Class IIIA), MAPC can form large micelles in aqueous media (Small, 1968). Soybean MAPC effectively enhances emulsion stability when being combined with nonionic surfactants such as glyceryl monostearate and sorbitane monostearate (Fujita et al., 1993). MAPC also promoted the permeability of 1-deamino-8-D-arginine-vasopressin when being tested in a Caco-2 cell monolayer (Hovgaard et al., 1995) and that of dextran and bovine serum albumin when being tested in a rat model (Tagesson et al., 1985). Using soybean MAPC in SNEDDS as a surfactant could be of special interest, when aiming to reduce the concentration of synthetic surfactants while maintaining adequate emulsification capacity, but, to the knowledge of the authors, this has not been investigated yet. The aims of the current study were therefore (1) to investigate the formulation of SNEDDS containing soybean MAPC using a design of experiments (DoE) approach and (2) to characterize SNEDDS in terms of dispersion properties, i.e. the emulsion droplet size and zeta potential in biorelevant media; in vitro lipolysis profile; and viscometric properties, when gradually replacing the synthetic surfactant Labrasol by MAPC. 2. Materials and methods 2.1. Materials Lipoid S LPC 80 (LPC) (from soybean, containing 80.8% MAPC and 13.2% phosphatidylcholine (PC)) and Lipoid S PC (from soybean, containing 98.0% PC) were kindly provided by Lipoid GmbH (Ludwigshafen am Rhein, Germany). Soybean oil, sodium taurodeoxycholate hydrate (NaTDC) (>95% pure), 2-(N-morpholino) ethanesulfonic acid (MES) hydrate, MES sodium salt, Trizma maleate, and pancreatin from porcine pancreas were purchased from Sigma-Aldrich (St. Louis, MO, USA). Maisine 35-1 (Maisine) (glyceryl monolinoleate) and Labrasol (caprylocaproyl macrogol8 glycerides) were gifts from Gattefossé (Saint-Priest, France). Captex 300 (Captex) (glyceryl tricaprylate/tricaprate) and Capmul MCM EP (Capmul) (glyceryl monocaprylate) were kindly provided by Abitec (Columbus, OH, USA). Absolute ethanol (99.9%) and sodium chloride were obtained from VWR (Radnor, PA, USA). Water was purified by a SG Ultraclear water system (SG Water GmbH, Barsbüttel, Germany). 2.2. Methods 2.2.1. Design of experiments A D-optimal experimental design of 13 experiments, including 3 center points, was generated by MODDE 10.1.0 software (Umetrics, Sweden) for two formulation sets containing 40% mixed glycerides, 30–55% Labrasol, 0–25% LPC, and 0–10% ethanol. Table 1 shows the composition of the DoE formulations. A classical

mixture design was not chosen because Labrasol, LPC and ethanol ranges are different (Eriksson et al., 2008). One formulation set was based on long-chain (LC) glycerides (soybean oil:Maisine (1:1 w/ w)), called “LC formulation set” hereafter, while the other set was based on medium-chain (MC) glycerides (Capmul:Captex (1:1 w/ w)), and was called “MC formulation set” hereafter. All formulations were dispersed in a fasted-state median intestinal fluid (FaMIF) (Madsen et al. 2016) (Table 2) prepared based on the median values of pH, bile salts and phospholipid concentrations, osmolarity and buffer capacity of human fasted-state intestinal fluids, as described by Bergstrom et al. (2014). The droplet size of the obtained emulsions was measured using dynamic light scattering (DLS) (described below). The matrix of the Labrasol, LPC and ethanol concentrations was correlated to the matrix of the droplet sizes by partial least square regression (Lundstedt et al., 1998) using the MODDE 10.1.0 software. The model quality was evaluated based on the goodness of fit (R2) and goodness of prediction (Q2). A R2 close to 1 refers to a model of good fit; a Q2 close to 1 refers to excellent prediction power; and Q2 larger than 0.5 refers to good prediction power (Eriksson et al., 2008). 2.2.2. Characterization of SNEDDS containing LPC 2.2.2.1. Droplet size and zeta-potential measurements. The investigated dispersions were prepared by adding the formulations to the biorelevant media at a ratio of 1:200 under gentle overhead stirring using a rotator (Intelli-Mixer RM-2M, In Vitro, Denmark) at 20 rpm and 37  C for 5 min. The biorelevant media used were the fasted-state simulated gastric fluid using NaTDC (FaSSGFTDC) (instead of using sodium taurocholate as in the original FaSSGF (Galia et al., 1998)) or FaMIF. Table 2 shows the composition of FaSSGFTDC and FaMIF. Preliminary experiments showed that pepsin at the concentration of 0.1 mg/mL did not influence the droplet size and zeta potential of the dispersion, but interfered with the measurement quality by making the medium turbid. Pepsin was therefore eliminated from the FaSSGFTDC used. The droplet size of the dispersion was measured by DLS and the zeta potential was measured by the mixed-mode measurement phase analysis light scattering technology (M3-PALS) at 37  C using a Zetasizer Nano ZS (Malvern, UK). The effect of pH on the droplet size and zeta potential was investigated by measuring droplet size and zeta potential of the nanoemulsions formed upon dispersion (1:200) in FaSSGFTDC or FaMIF at pH 1.6, 3.0, 5.0, 6.6 and 9.0. For all measurements, three independent samples of each formulation were investigated. 2.2.2.2. Microscopy studies. The systems formed when dispersing the formulations from the DoE in FaMIF were observed by optical microscopy at a magnification of 400(Zeiss Axiolab, Carl Zeiss GmbH, Germany) and cryogenic transmission electron microscopy (Cryo-TEM). For Cryo-TEM observation, 3 mL of the samples were applied on Lacy 3000 holey carbon film grids (Ted Pella Inc., CA, US). The grids were blotted in a Vitrobot automated vitrification device (FEI, Eindhoven, The Netherlands) under controlled environmental conditions (25  C, 100% relative humidity), then automatically plunged into liquid ethane to rapidly freeze the samples and transferred to liquid nitrogen (approximately 174  C). The frozen samples were then transferred to a Gatan 626 cryoholder (Gatan Inc., Warrendale, PA, USA) coupled to a FEI

Fig. 1. Chemical structure of MAPC.

T. Tran et al. / International Journal of Pharmaceutics 502 (2016) 151–160

153

Table 1 Composition of long-chain and medium-chain formulation sets established based on a D-optimal design. Experiment number

Composition (% w/w) LC Formulations MC formulations

1 2 3 4 5 6 7 8 9 10 11 12 13

Soybean oil Captex 300

Maisine 35-1 Capmul MCM EP

Labrasol Labrasol

Lipoid S LPC 80 Lipoid S LPC 80

Ethanol Ethanol

20.0 20.0 20.0 20.0 20.0 20.0 20.0 20.0 20.0 20.0 20.0 20.0 20.0

20.0 20.0 20.0 20.0 20.0 20.0 20.0 20.0 20.0 20.0 20.0 20.0 20.0

55.0 35.0 50.0 30.0 53.3 41.7 30.0 55.0 31.7 43.3 42.5 42.5 42.5

5.0 25.0 – 20.0 – 18.3 21.7 1.7 25.0 6.7 12.5 12.5 12.5

– – 10.0 10.0 6.7 – 8.3 3.3 3.3 10.0 5.0 5.0 5.0

Abbreviations: LC formulations: long-chain glycerides-containing formulations, MC formulations: medium-chain glycerides-containing formulations.

Tecnai G2 transmission electron microscope under low-dose conditions at 174  C (FEI, Eindhoven, The Netherlands). Images were recorded by a FEI Eager 4k CCD camera (FEI, Eindhoven, The Netherlands). 2.2.2.3. In vitro lipolysis. The experimental set-up (Metrohm AG, Herisau, Switzerland) consisted of a pH-stat apparatus, comprising a Titrando 842, a 804 Ti Stand, a 802 stirrer, a glass pH electrode and two 800 Dosino dosing units coupled to two 10 mL autoburettes. The apparatus was operated using the Tiamo 2.0 software (Metrohm AG, Herisau, Switzerland). Before adding pancreatic extract, 1 g or 0.7 g formulation was predispersed in 36 mL digestion medium (containing 2.95 mM NaTDC, 0.26 mM PC, 2.0 mM Trizma maleate, 73.0 mM sodium chloride, 1.4 mM calcium chloride, adjusted to pH 6.5) for 10 min in a thermostated vessel maintained at 37  C. The pH of the dispersion was manually adjusted to 6.5. The addition of 4 mL fresh pancreatic lipase extract (pH 6.5) (prepared as described by Williams et al. (2012)) to obtain an enzyme concentration of 400 USP/mL initiated the lipolysis reaction. The pH was maintained at 6.5 by adding NaOH solution for 60 min, and then rapidly raised to 9.0 using 1.0 M NaOH (backtitration process) to ionize all liberated unionized fatty acid (FA) (Williams et al., 2012). To investigate LPC digestion, a similar procedure was performed on a dispersion of 0.3 g LPC dispersed in 36 mL digestion medium (i.e. 8.33 mg/mL), prepared by stirring LPC in the medium overnight at 700 rpm until a homogenous translucent dispersion was obtained. All data of titrated FA presented in the result section have been corrected for the background data, i.e. the amount of FA released from the lipolysis of digestion medium (without any other lipid substrate).

Table 2 Composition of biorelevant media used to disperse the formulations. Components

NaTDC PC MES hydrate MES sodium salt Sodium chloride Sodium hydroxide (1.0 M) or Hydrochloric acid (1.0 M)

Concentration (mM) FaSSGFTDC (Galia et al., 1998)

FaMIF (Madsen et al., 2016)

0.08 0.02 – – 34.2 qs to pH 1.6  0.1

2.63 0.23 3.25 11.50 109.75 qs to pH 6.6  0.1

Abbreviations: FaSSGFTDC: fasted-state simulated gastric fluid, FaMIF: fasted-state simulated median intestinal fluids, MES: 2-(N-morpholino) ethanesulfonic acid, PC: phosphatidylcholine, NaTDC: sodium taurodeoxycholate.

2.2.2.4. Viscometric studies. Viscometric properties of SNEDDS were characterized using a AR-G2 rheometer with a stainless steel cone (40 mm diameter, 1 ) (TA Instruments, New Castle, DE, USA). A temperature ramp measurement was performed to determine the viscosity of the formulations at various temperature conditions when applying a shear rate of 100 s 1. The temperature was increased at the rate of 2  C/min, from 20  C to 70  C. In steady state flow studies, the shear stress of the formulation was measured at 20  C and 70  C as a function of the shear rate, which was increased from 0.001 to 1000 s 1. All samples were characterized in triplicate (independent samples). 3. Results 3.1. Partial least square regression model to predict nanoemulsion droplet sizes 3.1.1. Model fitting Formulations based on the DoE (Table 1) generated homogenous dispersions with different mean droplet sizes, depending on the types of the glycerides and the concentration of the surfactants and cosolvent used. The dispersion time was less than 2 min for all samples. All LC formulations generated turbid and polydisperse emulsions for which the droplet sizes could not be determined by DLS. This suggested a high polydispersity of the emulsions formed from all LC formulations within the investigated ranges. In contrast, all MC formulations generated nanoemulsions with droplet sizes ranging between 23 and 167 nm. Therefore, quantitative modeling of the droplet sizes was only applicable for the MC formulation set. A partial least square model with Q2 and R2 closest to 1 was selected, with square transformation on the data set, showing a good fit to experimental values with R2 of 0.97 and a good prediction power with Q2 of 0.82. The measured droplet sizes of random formulations (Table 3) in FaMIF are compared to the values predicted by the model to assess the model prediction power. The experimentally determined and predicted droplet size values are plotted against each other in Fig. 2. The data were fitted to the quadratic equation: y = b0 + b1x1 + b2x2 + b3x3 + b11x12 + b22x22 + b33x32 + b12x1x2 + b13x1x3 + b23x2x3 + e,

(1)

where the response y is the transformed value of the nanoemulsion droplet size, the three factors x1, x2, x3 are the fractions of Labrasol, LPC and ethanol, respectively, expressed in the 0–1 range, b0, b1, b2, b3, b11, b22, b33, b12, b13, b23 are equation coefficients, and e is the residual response variation. The coefficients of factors,

154

T. Tran et al. / International Journal of Pharmaceutics 502 (2016) 151–160

Table 3 Composition of random formulations used to test the prediction power of the model. Experiment number

14 15 16

Composition (% w/w) Captex 300

Capmul MCM EP

Labrasol

Lipoid S LPC 80

Ethanol

20 20 20

20 20 20

50 50 40

5 10 20

5 0 0

which are centered and scaled to unit variance, are displayed in the bar chart in Fig. 3a. These coefficients combined with their confidence intervals represented by error bars provide information about the statistical significance and the positive or negative effect of each equation term on the response value. The statistically insignificant terms are b3x3, b33x32 and b13x1x3, which pertain to the insignificant effect of ethanol concentration and the Labrasolethanol interaction to the droplet size. Further, the coefficient plot shows that the Labrasol concentration and the LPC square term contributed significantly to the increase of nanoemulsion droplet size, while the LPC concentration, the Labrasol-LPC and the LPCethanol interactions had a significant effect on reducing nanoemulsion droplet size. In order to evaluate the importance of each equation term, the variable importance in the projection (VIP) plot given by the MODDE software was studied (Fig. 3b). Equation terms with VIP values higher than 1.0 are considered to significantly influence the response value while ones with VIP values lower than 0.7–0.8 are considered unimportant (Eriksson et al., 2008; Wold et al., 1993). The surfactant-related factors, i.e. LPC concentration, Labrasol concentration, LPC concentration square term and Labrasol-LPC interaction term were the most influential factors affecting the nanoemulsion droplet size. The impact of ethanol was minor and based principally on the interactions with the two surfactants Labrasol and LPC. The droplet size screening and modeling provided a droplet size prediction plot for the MC formulation set (Fig. 4). All MC formulations in the investigated range were predicted to generate nanoemulsions in FaMIF. Increasing LPC and reducing Labrasol concentration in MC formulations shifted the obtained nanoemulsions to the area of smaller droplet sizes in the predicted ternary diagram. 3.1.2. Prediction of surfactant and cosolvent effects on nanoemulsion droplet sizes A factor effect plot predicted by the model is shown in Fig. 5. To create the plot, MODDE software determines a standard reference

mixture, which is the centroid of the constrained region, containing 40% glycerides, 42.5% Labrasol, 12.5% LPC and 5% ethanol. The plot presents the predicted droplet size when each factor is varied from its lowest to its highest level, the glyceride mixture is fixed at 40% and the ratios of the other two factors are kept as in the reference mixture. The factor effect plot shows the effect of each individual surfactant and the cosolvent on the emulsion droplet size. Using more than 35% Labrasol is predicted to increase the nanoemulsion droplet size. Increasing LPC from 0 to 20% is predicted to initially decrease the nanoemulsion droplet size, but using more than 20% LPC might increase the droplet size. The ethanol concentration is estimated to have an insignificant effect on the droplet size. 3.1.3. Microscopy study on dispersion of DoE formulations in FaMIF Optical and cryo-TEM microscopy studies were done to confirm the structures formed after dispersing the formulations in FaMIF. Under the light and electron microscopes, the systems obtained when dispersing LC and MC formulations from the established DoE in FaMIF were emulsions, but with different droplet size distributions. From the LC formulations, droplet sizes from nanoto micrometer-range were observed, indicating a large distribution range of droplet sizes (Fig. 6a and b). The investigated LC formulations were thus self-emulsifying, but not self-nanoemulsifying systems. In contrast, from the MC formulations, the cryoTEM study showed that nanoemulsion droplets were the predominant particle species in the obtained dispersions (Fig. 6c), while oil droplets were hardly found under the optical microscope. 3.2. Characterization of MC SNEDDS containing LPC Based on the DoE screening, LC formulations were excluded from further characterization due to lack of nanoemulsification capacity. In the following, MC SNEDDS were investigated to study the effect of LPC on the nanoemulsion droplet size, zeta potential,

Fig. 2. Experimentally determined versus predicted droplet sizes of the dispersions in FaMIF of 13 DoE MC SNEDDS (Table 1) (green symbols) and 3 random formulations prepared to test the predictive power of the model (Table 3) (orange symbols, with error bars connecting the lower and upper value predicted and measured). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

T. Tran et al. / International Journal of Pharmaceutics 502 (2016) 151–160

155

Fig. 3. (a) Centered and scaled coefficients corresponding to nine equation terms in Eq. (1). The asterisk (*) signifies coefficients relating to equation terms of insignificant effect on the response value. (b) Variable importance in the projection (VIP) plot with equation terms classed according to their importance to the response value. Equation terms with VIP-value higher than 1.0, i.e. LPC, Labrasol, and LPC square, are considered influential to the response value. x1, x2, x3 are the fractions of Labrasol, LPC and ethanol, respectively.

the in vitro digestion of the dispersed formulations and the viscometric properties of the formulations. Formulations F0 to F35 (the number corresponds to LPC concentration) containing 40% MC glycerides, 60–25% Labrasol, 0–35% LPC and F10*, F25*, F30* (LPCfree formulations containing the same MC glycerides:Labrasol ratio as F10, F25, and F30 respectively) were selected (Table 4). Ethanol was excluded from MC SNEDDS due to insignificant effect of ethanol on the formulation emulsification capacity.

3.2.1. Droplet size and zeta potential of emulsions formed after dispersing SNEDDS in FaSSGFTDC and FaMIF The appearance, droplet size and zeta potential of the emulsions from F0 to F35 in FaSSGFTDC and FaMIF are shown in Fig. 7. All resulting emulsions in FaSSGFTDC had zeta potentials ranging from 6 to +10 mV (Fig. 7a). Adding LPC increased the zeta potential of the oil droplets in FaSSGFTDC. LPC concentration had

Fig. 4. Prediction plot of medium-chain formulation set containing 20% Captex 300 and 20% Capmul MCM EP. Excipient concentration is presented as fractions. Nanoemulsion droplet sizes: below 40 nm, 40–80 nm, 80–120 nm, Above 120 nm.

156

T. Tran et al. / International Journal of Pharmaceutics 502 (2016) 151–160

Fig. 5. Factor effect plots showing the effect of each component on the nanoemulsion droplet size from MC SNEDDS. The displayed droplet size is the predicted value obtained when varying one excipient concentration (Labrasol, LPC or ethanol) and maintaining the other excipients’ ratio as in a reference mixture (containing 42.5% Labrasol, 12.5% LPC, 5% ethanol). 40% glycerides is fixed for all formulations.

negligible effect on the zeta potentials of emulsions in FaMIF, which ranged between 25 to 28 mV. Replacing 50% of Labrasol by LPC resulted in smaller mean nanoemulsion droplet sizes in both FaSSGFTDC and FaMIF. SNEDDS containing 20% LPC formed the nanoemulsion of the smallest mean droplet size, i.e. 78  3 nm (n = 3) in FaSSGFTDC and 50  5 nm (n = 3) in FaMIF. Formulations without LPC or containing 30% LPC formed polydisperse emulsions in FaSSGFTDC. The droplet size measurement results by DLS showed that formulations containing more than 30% LPC formed polydisperse emulsions in both FaSSGFTDC and FaMIF. Nanoemulsions in FaMIF had smaller mean droplet sizes than nanoemulsions in FaSSGFTDC from the same SNEDDS. The difference was less pronounced for formulations containing 15–27.5% LPC compared to the other formulations. The differences between the two media, primarily the pH and the NaTDC/PC micelle concentrations, made SNEDDS disperse differently in the two conditions in terms of size and zeta potential of emulsion droplets. To analyze the effect of these factors, the droplet size and zeta potential of emulsions at different pH values ranging from 1.6 to 9.0 of both FaSSGFTDC and FaMIF were studied and are shown in Fig. 8. At low pH conditions, the zeta potential of the emulsions was more positive, especially for formulations containing LPC (Fig. 8a and c), except for formulations without LPC in FaSSGFTDC. The emulsion droplet sizes increased at low pH, and this was more pronounced for formulations without LPC than for formulations containing LPC (Fig. 8b and d). Formulations without LPC formed polydisperse emulsion in FaSSGFTDC at pH values from 1.6 to 9.0.

The emulsification properties of F0, F10, F20 and F25 suggested that replacing Labrasol by LPC made the nanoemulsion droplet sizes less dependent on the pH, but the zeta potential increased stronger at low pH than formulations without LPC (i.e. F0, F10*, F25*). Smaller droplet sizes were obtained from formulations containing LPC in both media compared to the ones from LPC-free formulations containing the same glycerides:Labrasol ratio (F10 and F25 versus F10* and F25* respectively). The emulsions from LPC-free formulations containing different glycerides:Labrasol ratios, i.e. F0, F10* and F25*, had similar zeta potentials and particle sizes in FaSSGFTDC or FaMIF. 3.2.2. In vitro lipolysis of LPC dispersion and SNEDDS formulations The in vitro digestion catalyzed by porcine pancreatin of LPC and the effect of LPC on the lipolysis of lipid formulations were studied. Fig. 9 presents the in vitro lipolysis of formulations F0, F20, F30, F30* and a homogenous dispersion of 0.3 g LPC in digestion medium. Digestion of a pure LPC dispersion over 60 min liberated approximately 6% of the theoretically present FA. The calculation was based on the assumption that each mole of MAPC releases one mole of fatty acid and each mole of PC releases two moles of fatty acid. In the in vitro lipolysis model, the emulsions obtained when dispersing F20 and F30 in digestion medium (1 g in 40 mL) had the droplet sizes of 246  2 nm and 198  3 nm, respectively. The emulsions formed from F0 and F30* were polydisperse and at micrometer-range sizes. The replacement of Labrasol by LPC in F0, F20, and F30 resulted in statistically significantly less FA released in

Fig. 6. (a) Optical microscopy and (b and c) cryo-TEM images of emulsions obtained from dispersing the formulations containing (a and b) 40% LC glycerides, 41.7% Labrasol and 18.3% LPC or (c) 40% MC glycerides, 41.7% Labrasol, and 18.3% LPC in FaMIF (ratio 1:200).

T. Tran et al. / International Journal of Pharmaceutics 502 (2016) 151–160 Table 4 Composition of formulations with increasing LPC/Labrasol ratio and three LPC-free formulations (% w/w). Formulations

Captex 300

Capmul MCM EP

F0 F5 F10 F15 F20 F25 F27.5 F30 F32.5 F35 F10* F25* F30*

20 20 20 20 20 20 20 20 20 20 22 26.5 28.6

20 20 20 20 20 20 20 20 20 20 22 26.5 28.6

Surfactants Labrasol

Lipoid S LPC 80

60 55 50 45 40 35 32.5 30 27.5 25 56 47 42.8

0 5 10 15 20 25 27.5 30 32.5 35 0 0 0

F10*, F25*, and F30* are LPC-free formulations containing the same glycerides: Labrasol ratio as F10, F25, and F30 respectively. 0.7 g of F30* contains the same amount of Labrasol and glycerides as 1 g of F30.

total (2.05  0.02 mmol reduced to 1.76  0.05 mmol when incorporating 30% LPC) (Fig. 9c). Lipolysis of 1 g F30 was compared to lipolysis of 0.7 g F30* (the same amounts of glycerides and Labrasol were present in both formulations), to observe the effect of LPC on the digestion of other formulation components. The lower amount of ionized FA and total FA liberated from F30 compared to F30* suggested that LPC inhibited the digestion of glycerides and Labrasol. 3.2.3. Viscometric characterization of SNEDDS formulations The viscometric properties of formulations with different LPC concentrations, i.e. F0, F10, F20, F30, F35 are presented in Fig. 10. The physical state and viscosity of the formulations at room temperature depended on the concentration of LPC incorporated. Formulations containing 25% or more LPC were semi-solid, whereas all formulations containing equal or less than 20% LPC

157

were transparent solutions at room temperature (Fig. 7c). Increasing LPC concentration resulted in higher viscosity for the formulations. Going from formulation F0 to formulation F35, the viscosity increased from 57  1 to 436  8 mPa.s at room temperature (25  C) when applying the shear rate of 100 s 1. When the temperature was increased from 20 to 70  C, formulation viscosity reduced markedly, about 6–10 times, depending on LPC concentration (Fig. 10a). The formulation containing 35% of LPC had a viscosity of 579  11 mPa s at 20  C and 59  2 mPa s at 70  C, while formulations without LPC had a much lower viscosity (71  2 mPa s at 20  C and 11 mPa s at 70  C, Fig. 10a). In addition, the linear relationship between shear stress and shear rate of all formulations at both 20 and 70  C (Fig. 10b and c) suggested that all formulations were Newtonian fluids. 4. Discussion 4.1. Using LPC as a natural lipophilic surfactant in LC and MC SNEDDS The difference between the dispersions from the LC and MC formulation sets, which contain identical excipient types and proportions showed a substantial effect of fatty acid chain length on the self-nanoemulsification capacity of the lipid formulations. All MC formulations self-nanoemulsified in FaMIF, while all investigated LC formulations formed polydisperse emulsions of which the droplet size could not be measured by DLS. These results agree with Thomas et al. (2012) who found smaller emulsion droplet sizes from MC SNEDDS (containing Captex, Capmul, Cremophor RH40 or Cremophor EL and ethanol) than from LC SNEDDS (containing soybean oil, Maisine, Cremophor RH40 or Cremophor EL and ethanol). Research on replacing Labrasol by other surfactants to combine with LPC and increase the nanoemulsification capacity is required to extend the application of LPC in LC SNEDDS formulation. The performed DoE allowed screening the ability of different combinations of MC glycerides, Labrasol and LPC to form SNEDDS.

Fig. 7. (a) Zeta potential and (b) droplet size of the obtained emulsions in FaSSGFTDC (orange symbols) and FaMIF (green symbols) from F0 to F35 (MC glycerides, 60–25% Labrasol, 0–35% LPC). Data are presented as mean  SD (n = 3). (c) Appearance of F0 to F35 and their corresponding emulsions in (d) FaSSGFTDC and (e) FaMIF. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

158

T. Tran et al. / International Journal of Pharmaceutics 502 (2016) 151–160

Fig. 8. (a and c) Zeta potential and (b and d) droplet size of formulations containing LPC (closed square symbols), i.e. F10, F20, F25, and formulations without LPC (open triangular symbols), i.e. F0, F10* and F25* in pH-modified (left) FaSSGFTDC and (right) FaMIF. The emulsion droplets for F0, F10* and F25* in FaSSGFTDC could not be determined by DLS due to high polydispersity. Data are presented as mean  SD (n = 3).

From both predicted and experimental results, LPC had a significant effect on reducing nanoemulsion droplet size from MC formulations dispersed in FaSSGFTDC and FaMIF, when using up to 30% LPC to replace Labrasol. A minimum nanoemulsion droplet size in FaSSGFTDC and FaSSIF was obtained by incorporating 20% LPC. While the glycerides:Labrasol ratio had only a minor effect on the emulsion droplet sizes and zeta potentials, adding LPC to formulations of identical glycerides:Labrasol ratio reduced the emulsion droplet sizes in both simulated gastric and intestinal media. Thus, using an optimized ratio of LPC and Labrasol as lipophilic and hydrophilic surfactants maximized the dispersion capacity of SNEDDS, and reduced the necessity to formulate SNEDDS with a high amount of Labrasol. Although the interaction between ethanol and LPC was a significant factor affecting the emulsion droplet size from MC SNEDDS according to the partial least square model, the overall effect of ethanol on the emulsion droplet size was not significant

(Fig. 5c). Using ethanol in SNEDDS has two main practical disadvantages: (1) the risk of losing solubilization capacity upon rapid dispersion may trigger the precipitation of active ingredients and reduce absorption, and (2) upon filling SNEDDS into gelatin capsules, the diffusion of ethanol through the gelatin capsule wall may provoke ethanol evaporation during capsule drying and alter the capsule properties (Gullapalli, 2010). It is therefore advantageous to use ethanol-free SNEDDS for liquid or semi-solid filled capsule applications. SNEDDS containing 25% or more LPC were semi-solid at room temperature. However, this physical state did not affect the emulsification time, which were shorter than 2 min for all investigated formulations. The viscosity of formulations containing 5–35% LPC was determined at a high shear rate condition of 100 s 1 at 20–70  C, which represents a possible temperature range during the formulation (Podczeck and Jones, 2004). The obtained viscosity values were within the range for a technically

Fig. 9. (a and b) In vitro lipolysis profiles of formulations F0, F20, F30, F30* and LPC showing the amount of fatty acid titrated by addition of NaOH over 60 min. (c) Amount of ionized and unionized fatty acid liberated over 60 min of in vitro lipolysis for 1 g F0, F20, and F30, and 0.7 g F30*. The amount of ionized fatty acid was calculated based on the amount of fatty acid titrated over 60 min and that of unionized fatty acid was calculated based on the amount of fatty acid titrated during the back-titration process (Williams et al., 2012). The total amounts of fatty acid released from F0, F20, F30, and F30* are significantly different with p < 0.001, except the ones between F20 and F30*. Data are presented as mean  SD (n = 3).

T. Tran et al. / International Journal of Pharmaceutics 502 (2016) 151–160

159

Fig. 10. Viscometric behavior of formulations containing different LPC concentrations. (a) Viscosity of formulations as a function of temperature at a shear rate of 100 s 1. (b and c) Relationship between shear stress and shear rate of the formulation at (b) 20  C and (c) 70  C. Data are presented as mean  SD (n = 3).

feasible capsule filling process, which is required to be from 100 to 25,000 mPa s (Podczeck and Jones, 2004), suggesting the feasibility of developing SNEDDS containing LPC for capsule production. 4.2. Self-nanoemulsifying capacity of MC SNEDDS containing LPC in simulated gastric and intestinal media Due to the presence of more NaTDC/PC micelles in FaMIF, LPCfree formulations formed nanoemulsions in FaMIF, but not in FaSSGFTDC. The higher concentration of the bile salt in FaMIF rendered the nanoemulsion droplet sizes smaller and the droplets more negatively charged. However, the presence of 15–27.5% LPC reduced the difference between the emulsion droplet sizes in FaSSGFTDC and FaMIF (Fig. 7). When adjusting the pH of FaMIF to 1.6, the sizes and zeta potentials of the formed emulsion droplets increased because of the protonation and possible precipitation of the bile salt. In addition, the zeta potentials of the emulsions containing LPC were more positive than formulations without LPC because the phosphate groups of MAPC and PC in LPC have a low pKa (from 0 to 2) (Hanahan, 1997), making MAPC and PC partly positively charged at low pH. Therefore, the dependence of zeta potential values on the pH was more pronounced for emulsions containing a higher concentration of LPC (Fig. 8a and c). However, the emulsion droplet sizes of formulations containing LPC were less sensitive to the pH change of the media when compared to formulations without LPC (Fig. 8b and d). 4.3. In vitro lipolysis of SNEDDS containing LPC SNEDDS containing the same amount of glycerides but different Labrasol and LPC concentrations were digested in the in vitro lipolysis model. The droplet size of emulsions in the digestion medium could be ranked as follows: F30 < F20 < F0, suggesting the interfacial area for lipase-mediated digestion and the digestion rate should be ranked as F30 > F20 > F0 because lipolysis driven by lipase is an interfacial process (Golding and Wooster, 2010). However, the digestion rate and extent of digestion of the investigated formulations were ranked as F0 > F20 > F30, indicating a more pronounced dependence of the digestion profiles on the LPC concentration than on the nanoemulsion droplet size. Increasing LPC concentrations in the SNEDDS reduced the rate and extent of digestion. The digestion of 0.7 g F30* liberated a higher total amount of FA compared to the digestion of the same amount of lipid substrate combined with 0.3 g LPC (i.e. 1 g F30) suggesting a digestion inhibiting effect of LPC in F30 (30% LPC). Therefore, using LPC partly inhibited the digestion of glycerides and Labrasol. Since an

inaccessibility of oil/water interface reduces the rate and extent of digestion (Golding and Wooster, 2010), it can be speculated that LPC was presented at the oil/water interface and only digested at a low extent, thereby inhibited the adsorption of the pancreatic colipase/lipase complex to the emulsified lipid droplet and in turn the rate of digestion. A disadvantage of some SNEDDS is the possible loss of drug solubilization capacity upon rapid lipid digestion, producing undesirable precipitation of poorly watersoluble drugs and an insufficient intestinal absorption (Porter et al., 2007). The inhibitory effect of LPC might be useful to modify the digestion of the formulation, prolong the solubilization of drugs and avoid drug precipitation. Therefore, further investigations using drug-loaded SNEDDS containing LPC should be carried out. 5. Conclusions The work presented in this study demonstrates the feasibility of using MAPC as a lipophilic surfactant in MC SNEDDS to reduce the synthetic surfactant and cosolvent concentrations and to enhance the capacity of SNEDDS to form nanoemulsions of small droplet sizes. The choice of MAPC concentration affected the emulsion droplet sizes and was crucial for the medium-dependence of the droplet sizes. 20% LPC (i.e. 16.1% MAPC) combined with 40% Labrasol was the combination that formed nanoemulsions of smallest droplet size (lower than 80 nm in both simulated gastric and intestinal media). Although MAPC is positively charged at low pH making the nanoemulsion zeta potential more positive, the droplet size of the SNEDDS dispersions remained relatively unaffected in the entire gastrointestinal pH range. Furthermore, the study provided evidence that incorporating MAPC inhibited the digestion of the glycerides and Labrasol and increased the dynamic viscosity of the formulations. Further studies investigating the formulation of MC and LC SNEDDS containing MAPC with other surfactants than Labrasol and the effect of MAPC on drug solubilization and absorption from SNEDDS are required to gain further insights into MAPC utilization for SNEDDS development. Funding sources The research was supported by the University of Copenhagen and a grant from the Phospholipid Research Center Heidelberg, Germany. Acknowledgments Support from the University of Copenhagen and the Phospholipid Research Center Heidelberg through the research grant is kindly acknowledged. We also thank Ramon Liebrechts from the

160

T. Tran et al. / International Journal of Pharmaceutics 502 (2016) 151–160

Core Facility for Integrated Microscopy, University of Copenhagen for the support with cryo-TEM imaging. References Attwood, D., Florence, A.T., 1983. Aspects of surfactant toxicity, Surfactant Systems: Their Chemistry, Pharmacy and Biology. 1st ed. Chapman and Hall Ltd., London, pp. 617–697. Bergstrom, C.A., Holm, R., Jorgensen, S.A., Andersson, S.B., Artursson, P., Beato, S., Borde, A., Box, K., Brewster, M., Dressman, J., Feng, K.I., Halbert, G., Kostewicz, E., McAllister, M., Muenster, U., Thinnes, J., Taylor, R., Mullertz, A., 2014. Early pharmaceutical profiling to predict oral drug absorption: current status and unmet needs. Eur. J. Pharm. Sci. 57, 173–199. Buyukozturk, F., Benneyan, J.C., Carrier, R.L., 2010. Impact of emulsion-based drug delivery systems on intestinal permeability and drug release kinetics. J. Control. Release 142 (1), 22–30. Caliph, S.M., Charman, W.N., Porter, C.J., 2000. Effect of short-, medium-, and longchain fatty acid-based vehicles on the absolute oral bioavailability and intestinal lymphatic transport of halofantrine and assessment of mass balance in lymphcannulated and non-cannulated rats. J. Pharm Sci. 89 (8), 1073–1084. Carey, M.C., Small, D.M., Bliss, C.M., 1983. Lipid digestion and absorption. Annu. Rev. Physiol. 45 (1), 651–677. Constantinides, P.P., Wasan, K.M., 2007. Lipid formulation strategies for enhancing intestinal transport and absorption of P-glycoprotein (P-gp) substrate drugs: in vitro/in vivo case studies. J. Pharm. Sci. 96 (2), 235–248. Eriksson, L., Johansson, E., Kettaneh-Wold, N., Wikström, C., Wold, S., 2008. Design of Experiments: Principles and Applications. MKS Umetrics AB. Fujita, S., Suzuki, A., Yahisa, E., 1993. Enhanced emulsifying ability of food surfactants by addition of lysophospholipids. Food Hydrocolloids: Structures, Properties, and Functions. Springer, Dordrech, pp. 429–433. Galia, E., Nicolaides, E., Hörter, D., Löbenberg, R., Reppas, C., Dressman, J., 1998. Evaluation of various dissolution media for predicting in vivo performance of class I and II drugs. Pharm. Res. 15 (5), 698–705. Gloxhuber, C., 1974. Toxicological properties of surfactants. Arch. Toxicol. 32 (4), 245–269. Golding, M., Wooster, T.J., 2010. The influence of emulsion structure and stability on lipid digestion. Curr. Opin. Colloid Interface Sci. 15 (1–2), 90–101. Gullapalli, R.P., 2010. Soft gelatin capsules (softgels). J. Pharm. Sci. 99 (10), 4107– 4148. Gupta, S., Chavhan, S., Sawant, K.K., 2011. Self-nanoemulsifying drug delivery system for adefovir dipivoxil: design, characterization, in vitro and ex vivo evaluation. Colloids Surf. A 392 (1), 145–155. Hanahan, D.J., 1997. Choline-containing phospholipids: diacyl-, alkylacyl-, and alkenylacylcholine phosphoglycerides and sphingomyelin. A Guide to Phospholipid Chemistry. Oxford University Press, pp. 61–130. Hovgaard, L., Brøndsted, H., Nielsen, H.M., 1995. Drug delivery studies in Caco2 monolayers. II. Absorption enhancer effects of lysophosphatidylcholines. Int. J. Pharm. 114 (2), 141–149.

Kovarik, J.M., Mueller, E.A., van Bree, J.B., Tetzloff, W., Kutz, K., 1994. Reduced interand intraindividual variability in cyclosporine pharmacokinetics from a microemulsion formulation. J. Pharm. Sci. 83 (3), 444–446. Larsen, A.T., Ohlsson, A.G., Polentarutti, B., Barker, R.A., Phillips, A.R., Abu-Rmaileh, R., Dickinson, P.A., Abrahamsson, B., Ostergaard, J., Mullertz, A., 2013. Oral bioavailability of cinnarizine in dogs: relation to SNEDDS droplet size, drug solubility and in vitro precipitation. Eur. J. Pharm Sci. 48 (1–2), 339–350. Lipinski, C.A., Lombardo, F., Dominy, B.W., Feeney, P.J., 2012. Experimental and computational approaches to estimate solubility and permeability in drug discovery and development settings. Adv. Drug Deliv. Rev. 64, 4–17. Lipinski, C.A., 2000. Drug-like properties and the causes of poor solubility and poor permeability. J. Pharm. Toxicol. Methods 44 (1), 235–249. Lundstedt, T., Seifert, E., Abramo, L., Thelin, B., Nyström, Å., Pettersen, J., Bergman, R., 1998. Experimental design and optimization. Chemometr. Intell. Lab. Syst. 42 (1), 3–40. Madsen, C.M., Boyd, B., Müllertz, A., Rades, T., 2016 Supersaturation of zafirlukast in fasted and fed state intestinal media with and without precipitation inhibitors. Unpublished manuscript. Nielsen, F.S., Petersen, K.B., Mullertz, A., 2008. Bioavailability of probucol from lipid and surfactant based formulations in minipigs: influence of droplet size and dietary state. Eur. J. Pharm. Biopharm. 69 (2), 553–562. Patel, J.P., Brocks, D.R., 2009. The effect of oral lipids and circulating lipoproteins on the metabolism of drugs. Expert Opin. Drug Metab. Toxicol. 5 (11), 1385–1398. Podczeck, F., Jones, B.E., 2004. Pharmaceutical Capsules. Pharmaceutical Press. Porter, C.J., Trevaskis, N.L., Charman, W.N., 2007. Lipids and lipid-based formulations: optimizing the oral delivery of lipophilic drugs. Nat. Rev. Drug Discov. 6 (3), 231–248. Pouton, C.W., 2000. Lipid formulations for oral administration of drugs: nonemulsifying, self-emulsifying and ‘self-microemulsifying’ drug delivery systems. Eur. J. Pharm. Sci. 11 (Suppl. 2), S93–S98. Small, D., 1968. A classification of biologic lipids based upon their interaction in aqueous systems. J. Am. Oil Chem. Soc. 45 (3), 108–119. Tagesson, C., Franzen, L., Dahl, G., Westrom, B., 1985. Lysophosphatidylcholine increases rat ileal permeability to macromolecules. Gut 26 (4), 369–377. Thomas, N., Mullertz, A., Graf, A., Rades, T., 2012. Influence of lipid composition and drug load on the in vitro performance of self-nanoemulsifying drug delivery systems. J. Pharm. Sci. 101 (5), 1721–1731. Tso, P., 1985. Gastrointestinal digestion and absorption of lipid. Adv. Lipid Res. 21, 143–186. Van Citters, G., Lin, H., 1999. The ileal brake: a fifteen-year progress report. Curr. Gastroenterol. Rep. 1 (5), 404–409. Williams, H.D., Sassene, P., Kleberg, K., Bakala-N'Goma, J.C., Calderone, M., Jannin, V., Igonin, A., Partheil, A., Marchaud, D., Jule, E., Vertommen, J., Maio, M., Blundell, R., Benameur, H., Carriere, F., Mullertz, A., Porter, C.J., Pouton, C.W., 2012. Toward the establishment of standardized in vitro tests for lipid-based formulations, part 1: method parameterization and comparison of in vitro digestion profiles across a range of representative formulations. J. Pharm. Sci. 101 (9), 3360–3380. Wold, S., Johansson, E., Cocchi, M., 1993. PLS-Partial least-square projections to latent structures. In: Kubinyi, H. (Ed.), 3D QSAR in Drug Design: Volume 1: Theory Methods and Applications. Springer Science & Business Media, pp. 523–550.