- Email: [email protected]
Dialkylphosphatidylcholine and egg yolk lecithin for emulsification of various triglycerides
Colloids and Surfaces B: Biointerfaces 41 (2005) 305–311
Dialkylphosphatidylcholine and egg yolk lecithin for emulsification of various triglycerides Tomoko Nii, Fumiyoshi Ishii ∗ Department of Pharmaceutical Sciences, Meiji Pharmaceutical University, 2-522-1 Noshio, Kiyose, Tokyo 204-8588, Japan Received 27 October 2004; accepted 23 December 2004
Abstract Synthesized saturated phosphatidylcholine (PC) and egg yolk lecithin (EYL) were investigated to explore their influence on particle sizes in emulsions when dispersing various triglycerides (TG). One of four different kinds of synthesized saturated PC (DLPC, DMPC, DPPC and DSPC) or three different kinds of EYL (purified EYL (PEL) and hydrogenated purified EYL with two different iodine values (IV), R-20 and R-5), 2.5% (w/w) glycerol solution and one of four kinds of TG (tricaprylin, tricaprin, trilaurin and trimyristin) were sonicated five times for 1 min with intervals of 0.5 min. When using four kinds of synthesized saturated PCs as emulsifiers, the carbon numbers of each PC had a strong correlation with the mean diameters of the emulsion when analyzed with each of the four kinds of TG used in the study (regression function ranged from 0.811 to 0.915). The carbon numbers of the TG had less correlation with the mean diameters than the PC in simple regression analysis (regression function ranged from 0.236 to 0.875). Multiple regression analysis using the carbon numbers both of the PC and TG as independent variables was remarkably significant in the regression function (2.0 × 10−14 ) and all regression coefficients (2.7 × 10−13 , 5.8 × 10−7 and 1.9 × 10−9 for PC, TG and intercept, respectively). Among the regression coefficients, the contribution of the carbon number of the PC was the most significant. These results indicated that a multiple regression function should be useful to estimate the mean diameters of emulsion droplets in any combinations of PC and TG used in this study. In the experiments using three kinds of EYL, the mean diameters also tended to increase according to the order of PEL, R-20 and R-5, which corresponds to the order of degrees of saturation (IV = 75, 20 and 2, respectively). The experimental values for EYL were compared with the estimated values calculated by the multiple regression function derived from synthesized PC data using the arithmetic carbon number, based on the components of each EYL. The estimated mean diameters were at comparable levels to the corresponding experimental mean diameters in the most saturated hydrogenated lecithin (R-5), while those were larger than the experimental mean diameters in two less saturated kinds of lecithin (R-20 and purified EYL). These findings gave useful information on the mean diameters of emulsion droplets when designing an emulsion formulation using a particular combination of a phospholipid and triglyceride. © 2005 Elsevier B.V. All rights reserved. Keywords: Phosphatidylcholine; Purified egg yolk lecithin; Hydrogenated egg yolk lecithin; Emulsion; Triglyceride; Particle size; Hydrophilic–lipophilic balance
1. Introduction Submicron oil-in-water (O/W) emulsions are expected to be a practical and promising option to disperse drugs that have the problem of poor water solubility into an aqueous medium [1,2]. There are numerous variations in the emulsifier system, but the surfactants and oil phases commonly ∗
Corresponding author. Tel.: +81 424 95 8468; fax: +81 424 95 8468. E-mail address: [email protected] (F. Ishii).
0927-7765/$ – see front matter © 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.colsurfb.2004.12.017
used for parenteral administration are limited [3]. Lecithins are the most common surfactant [4–6], while vegetable oils or medium-chain triglycerides (MCT) are commonly used for the oil phase in parenteral emulsions [3]. In recent years, melt-emulsified nanoparticles based on lipids that are solid at room temperature have been more prevalent [1,7,8]. Such solid lipid nanoparticles (SLN) prepared with triglycerides (TGs) and phospholipids have the advantages of a longer retention of the incorporated drugs than lipid emulsions and a more rapid degradation in the body than polymeric
306
T. Nii, F. Ishii / Colloids and Surfaces B: Biointerfaces 41 (2005) 305–311
nanoparticles prepared with polymers [8]. The SLN bears many similarities to the lipid emulsion and also liposomes, for phospholipids are on the surface of the particulates in these colloidal systems [8]. For intravenous application, particle size is considered a major issue since the half-lives of SLN were found to be dependent on their size [9]. In addition, variations of the particle size within the colloidal range may affect the physicochemical properties of the nanoparticles [9]. Accordingly, the particle size of the colloidal lipid particles is an important characteristic and thus, potential factors influencing the particle size are of interest. We previously studied the characteristics of various saturated and unsaturated phosphatidylcholines (PCs) used for emulsifying a MCT [10,11]. Particle sizes were influenced by the length and degrees of unsaturation of the acyl hydrocarbon chains of the PC. We suggested that the physical length and/or the shape of hydrocarbon chains of the PC and hydrophilic–lipophilic balance (HLB) of the PC might influence the particle sizes of the emulsion [10]. The primary objective of the current study was to investigate the characteristics of synthesized saturated PCs that possibly influence the particle sizes of the emulsion using various triglycerides. The relationship between the hydrocarbon chains in the PC used as the emulsifier and the hydrocarbon chains in the TG as the oil phase is a matter of particular interest. In addition, purified egg yolk lecithin and its hydrogenated compounds were used as mixtures of various phospholipids to explore the differences from pure PCs when emulsifying various TGs.
2. Materials and methods 2.1. Materials The following synthesized saturated PCs were purchased from NOF Corporation (Tokyo, Japan): l-␣-phosphatidylcholine dilauroyl (DLPC), l-␣-phosphatidylcholine dimyristoyl (DMPC), l-␣-phosphatidylcholine dipalmitoyl (DPPC) and l-␣-phosphatidylcholine distearoyl (DSPC). Glycerol trioctanoate (tricaprylin), glycerol tridecanoate (tricaprin), glyceryl trilaurate (trilaurin) and glycerol trimyristate (trimyristin) were purchased from Tokyo Kasei Kogyo Co. Ltd. (Tokyo, Japan). Purified egg yolk lecithin (PEL, iodine value, IV = 75), hydrogenated purified egg yolk lecithin with two different iodine values, i.e., R-20 (IV = 20) and R-5 (IV = 2) were kindly provided by Asahi Chemical Industries Co. Ltd. (Tokyo, Japan). The compositions of the acyl hydrocarbon chains for these lecithins are shown in Table 1 [12]. The glycerol was of the highest grade, which was a product of Wako Pure Chemical Industries (Osaka, Japan). 2.2. Preparation of the emulsion The emulsion was prepared by probe sonication. Each component of the emulsion was weighed into a glass container according to the ratio, as previously reported
Table 1 Composition of acyl hydrocarbon chains in egg yolk lecithin [12] PEL
R-20
R-5
C-16 16:0 16:1
30.4 1.5
29.4 0.0
29.3 0.0
C-18 18:0 18:1 18:2 18:3
15.2 27.7 15.5 0.0
35.8 20.0 0.0 0.0
55.8 0.0 0.0 0.0
C-20 20:0 20:1 20:2 20:4
0.0 0.0 0.0 5.5
3.2 3.6 0.6 0.0
7.7 0.0 0.0 0.0
C-22 22:0 22:1 22:2 22:5 22:6
0.0 0.0 0.0 0.0 4.1
2.9 3.0 1.2 0.0 0.0
7.0 0.0 0.0 0.0 0.0
Others
0.1
0.2
0.2
Data represent relative amounts (%).
[10,11,13], i.e., one of four kinds of synthesized PC or three kinds of lecithin (120 mg), 2.5% (w/w) glycerol solution (9.0 g) and one of four kinds of triglyceride (1.0 g). The mixture was sonicated five times for 1 min with intervals of 0.5 min with a probe ultrasonicator (VP-30S, Taitec Corporation, Japan, 20 kHz). The sonication was conducted at 65 ◦ C, which was sufficiently above the phase transition temperatures of all phospholipids, as well as the melting points of all triglycerides used in this study. 2.3. Particle size measurement Mean diameters of droplets in O/W emulsions were measured by a submicron particle size analyzer (NICOMP, Model 370, Pacific Scientific Instrument Division Co. Ltd.). 2.4. Statistics A statistical analysis was performed to examine the correlation between the mean diameters and carbon numbers of the acyl hydrocarbon chains in synthesized PCs or TGs. The following numbers were used for analysis. Synthesized PC: DLPC; 12, DMPC; 14, DPPC; 16 and DSPC; 18. TG: tricaprylin; 8, tricaprin; 10, trilaurin; 12 and trimyristin; 14. The correlation between the carbon numbers and the mean diameters was analyzed using Spearman’s correlation coefficient with the rank test. Moreover, a regression analysis was performed to examine if the carbon numbers of the acyl
T. Nii, F. Ishii / Colloids and Surfaces B: Biointerfaces 41 (2005) 305–311
hydrocarbon chains in synthesized PCs or TGs were associated with the mean diameters.
3. Results 3.1. Emulsion using synthesized saturated PC The results for the emulsion using synthesized PCs are presented in Figs. 1 and 2. Fig. 1 has the data grouped according to the TGs used for the oil phase. With all TGs, the mean diameters of the emulsion droplets clearly increased as the number of carbons in the acyl hydrocarbon chains in the PC increased. The same data grouped according to the PCs used as the emulsifier are shown in Fig. 2. The mean diameters also tended to increase as the number of carbons
Fig. 1. Mean diameters of emulsion droplets using four kinds of TGs with different hydrocarbon chain lengths (mean ± S.D., n = 3). Each TG was emulsified with four kinds of synthesized PCs with different hydrocarbon chain lengths.
Fig. 2. Mean diameters of emulsion droplets using four kinds of synthesized PCs with different hydrocarbon chain lengths (mean ± S.D., n = 3). Each PC was used to emulsify four kinds of TGs with different hydrocarbon chain lengths.
307
in the acyl hydrocarbon chains of TG increased. The difference in the particle sizes among the TGs was less clear in the shorter chain PCs (DLPC, DMPC), e.g., the mean diameters in tricaprylin, tricaprin, trilaurin and trimyristin were 116.4, 124.0, 144.3 and 147.0 in DLPC, respectively. In contrast, the particle sizes with the longer chain PCs (DPPC, DSPC) were more dependent on the number of carbons in the acyl hydrocarbon chains, i.e., the mean diameters of the above four TGs in DSPC were 204.4, 211.9, 335.5 and 390.8, respectively. With all PCs, to greater or lesser degrees, the particle sizes prepared with tricaprylin and tricaprin were similar while those prepared with trilaurin and trimyristin tended to increase more than these two shorter chain TGs. To ascertain whether the carbon numbers of the acyl hydrocarbon chains in the synthesized PCs or TGs were potential factors affecting the particle sizes of emulsion droplets, the correlation of the mean diameters and carbon numbers was statistically analyzed. Table 2 summarizes the coefficients and p-values of these analyses. The correlation coefficients and p-values obtained from Spearman’s correlation coefficient with the rank test are shown in the table. With regard to the regression analysis, the coefficient of determination and the p-values of the regression function tested by ANOVA were demonstrated to examine if the regression function was useful to estimate the mean diameters when using carbon numbers as explanatory variables. The significance of the regression coefficients, i.e., the intercept and independent variable(s), were also presented to examine if these factors were necessary to estimate the dependent variable. First, a simple regression analysis was performed for each TG (Group A, corresponding to Fig. 1) and each PC (Group B, corresponding to Fig. 2). In Group A, the carbon numbers of PC had a strong correlation with the mean diameter of emulsion when analyzed with each TG. The correlation coefficient (rs) in each TG ranged from 0.907 to 0.972, which meant that the mean diameters had a very strong correlation with the carbon number of the PC in every oil phase. Simple regression analysis also supported this strong correlation. The coefficient of determination in each TG was high, ranging from 0.811 to 0.915. The p-values of the regression lines were all below 0.0001, which was highly significant. These results demonstrated that the regression line of the carbon numbers of the PC was useful to estimate the mean diameter of the emulsion droplets. The regression coefficient was also significant in both its intercept and independent variable. However, the intercept was much less significant than the independent variable, which meant that the independent variables contributed more to the regression line. In Group B, the carbon numbers of the TG showed less of a correlation with the mean diameter than the PC. The correlation coefficient ranged from 0.561 to 0.907 and the coefficient of determination from 0.236 to 0.875. DMPC was not significant in Spearman’s correlation coefficient according to the rank test and regression analysis. It was notable that longer chain PCs
308
T. Nii, F. Ishii / Colloids and Surfaces B: Biointerfaces 41 (2005) 305–311
Table 2 Statistical analysis of correlation coefficient and regression analysis among particle sizes, carbon numbers of the phosphatidylcholine and carbon numbers of the triglyceride in emulsion using synthesized phosphatidylcholine Group
Independent variable 1
Independent variable 2
Synthesized phosphatidylcholine Group A: Each TG C number of PC (12–18) TG-C8 TG-C10 TG-C12 TG-C14
n
Correlation coefficienta
Regression analysisb
Correlation p coefficient
Regression function
Regression coefficient
Coefficient of determination
p
Intercept, p
Independent Independent variable 1, p variable 2, p
12 12 12 12
0.972 0.972 0.907 0.907
0.0013 0.0013 0.0026 0.0026
0.870 0.915 0.832 0.811
0.000006 0.000001 0.000022 0.000040
0.0260 0.0102 0.0026 0.0018
0.000006 0.000001 0.000022 0.000040
Group B: Each PC C number of TG (8–14) PC-C12 PC-C14 PC-C16 PC-C18
12 12 12 12
0.907 0.561 0.842 0.907
0.0026 0.0626 0.0052 0.0026
0.501 0.236 0.787 0.875
0.006012 0.062259 0.000073 0.000005
0.0028 0.0178 0.0212 0.0650
0.006012 0.062259 0.000073 0.000005
Group C: All sample Simple regression analysis C number of PC (12–18) C number of TG (8–14)
48 0.831 48 0.433
1.2 × 10−8 0.560 0.0030 0.167
5.9 × 10−10 0.002287
0.0004 0.5348
5.9 × 10−10 0.002287
0.743
2.0 × 10−14
1.9 × 10−9
2.7 × 10−13
Multiple regression analysis C number of PC (12–18)
C number of TG (8–14)
48
5.8 × 10−7
Abbreviations: TG-C8, tricaprylin; TG-C10, tricaprin; TG-C12, trilaurin; TG-C14, trimyristin; PC-C12, DLPC; PC-C14, DMPC; PC-C16, DPPC; PC-C18, DSPC. a Spearman’s correlation coefficient by rank test was used to examine correlation between the C numbers and mean diameters. b Groups A and B were examined by simple regression analysis while Group C was examined by both simple regression analysis and multiple regression analysis using the C numbers of the TG and PC of each sample.
(DPPC and DSPC) tended to indicate a stronger correlation with the carbon numbers of the TG than the shorter chain PCs. Next, all data were examined by simple regression and multiple regression analysis (Group C). In simple regression analysis, all data were tested using only a single independent variable, i.e., the carbon numbers of the PC or TG. A strong correlation was still maintained between the carbon numbers of the PC and the mean diameters, regardless of the TG in the oil phase, but the correlation coefficient and in particular the coefficient of determination were lowered. It was anticipated from the results of Group B that a correlation between the carbon numbers of the TG and the mean diameters was considerably lower, even though it was still significant. The correlation coefficient was 0.433 and the coefficient of determination was as low as 0.167, mainly due to the highly insignificant level of the intercept in the regression line. Multiple regression analysis was then performed using both the carbon numbers of the PC (variable 1) and TG (variable 2) as independent variables. The results of the analysis were remarkably significant, that is, the coefficient of determination was 0.743, for which the p-value was 2.0 × 10−14 . Concerning the regression coefficient, the p-values of variable 1, variable 2 and the intercept were 2.7 × 10−13 , 5.8 × 10−7 and 1.9 × 10−9 ,
respectively, and the contribution of the carbon numbers of the PC (variable 1) was most significant among these three regression coefficients. These results indicated that a multiple regression function might be useful to estimate the mean diameters of emulsion droplets in any combination of PC and TG used in this study. 3.2. Emulsion using egg yolk lecithin Figs. 3 and 4present the results of emulsion using egg yolk lecithin. Fig. 3 grouped the data by TG used for the oil phase. The mean diameters also tended to increase according to the order of PEL, R-20 and R-5, which corresponds to the order of degrees of saturation (IV = 75, 20 and 2, respectively). R20 fell between PEL and R-5, and the mean diameters of R-20 were closer to PEL in the shorter chain TGs while they were closer to R-5 in the longer chain TGs. Fig. 4 shows the same data grouped by the lecithin used as the emulsifier. For all lecithins, the mean diameters of the emulsion droplets increased as the number of carbons in the acyl hydrocarbon chains in TG increased. For egg yolk lecithin, the correlation between the carbon numbers of the TG and the mean diameters was statistically analyzed, and the results are summarized in Table 3. Simple
T. Nii, F. Ishii / Colloids and Surfaces B: Biointerfaces 41 (2005) 305–311
Fig. 3. Mean diameters of emulsion droplets using four kinds of TGs with different hydrocarbon chain lengths (mean ± S.D., n = 3). Each TG was emulsified with three kinds of egg yolk lecithins with different degrees of hydration.
regression analysis was performed for each lecithin (Group D, corresponding to Fig. 4). The carbon numbers of the TG indicated a significant correlation, but the degrees of significance were not as strong in PEL and R-5. The correlation coefficient was 0.756 and 0.713, and the coefficient of determination was 0.441 and 0.375 in PEL and R-5, respectively. For these lecithins, the contribution of the independent variable, i.e., the carbon numbers of the TG, to the regression line was rather weak. In contrast, R-20 showed a stronger correlation with the carbon numbers of the TG. The regression line was also highly significant, though the intercept contributed least to the regression line. Then, all data were examined by simple regression analysis. A significant correlation between
309
Fig. 4. Mean diameters of emulsion droplets using three kinds of egg yolk lecithins with different degrees of hydration (mean ± S.D., n = 3). Each egg yolk lecithin was used to emulsify four kinds of TGs with different hydrocarbon chain lengths.
the carbon numbers of the TG and the mean diameters was still evident. The correlation coefficient was 0.580 and the coefficient of determination was 0.313.
4. Discussion Particle sizes of O/W emulsions were influenced by various factors, for instance, dispersed oil and its ratio to the continuous water phase, the emulsifier and its concentration in total colloidal solution and the method of emulsion preparation [14]. In this study, various combinations of TGs and phospholipids were dispersed in glycerol solution by probe
Table 3 Statistical analysis of the correlation coefficient and regression analysis between particle sizes and carbon numbers of the triglyceride in emulsion using egg yolk lecithin Group
Independent variable 1
Independent n variable 2
Correlation coefficienta
Regression analysisb
Correlation p coefficient
Regression function
Regression coefficient
Coefficient of p determination
Intercept, p Independent variable 1, p
Egg yolk lecithin Group D: Each EYL PEL C number of TG (8–14) R-20 R-5
12 0.756 12 0.928 12 0.713
0.0122 0.0021 0.0181
0.441 0.807 0.375
0.011036 0.0001 0.000044 0.2958 0.020188 0.0015
0.011036 0.000044 0.020188
Group E: All sample Simple regression analysis C number of TG (8–14)
36 0.580
0.0006
0.313
0.000229 0.0009
0.000229
Independent variable 2, p
Abbreviations: EYL, egg yolk lecithin; PEL, purified egg yolk lecithin (IV = 75); R-20, hydrogenated purified egg yolk lecithin (IV = 20); R-5, hydrogenated purified egg yolk lecithin (IV = 2). a Spearman’s correlation coefficient by rank test was used to examine the correlation between C numbers and mean diameters. b Simple regression analysis was used in both Groups D and E.
310
T. Nii, F. Ishii / Colloids and Surfaces B: Biointerfaces 41 (2005) 305–311
sonication to investigate how their characteristics affected the particle sizes of the emulsion droplets. Concerning the synthesized PCs, the particle sizes of the emulsion droplets showed a strong correlation with the carbon numbers of the acyl hydrocarbon chains of PC when dispersing the same TG (rs = 0.907–0.972). In a previous study using tricaprylin, the mean diameters showed a strongly positive correlation with the carbon number of the PC when the emulsion was prepared at 65 ◦ C, which was sufficiently above the transition temperatures of all PCs used [10]. The results were consistent with our current study not only in tricaprylin but also other TGs having longer acyl hydrocarbon chains. It was reported that PCs with shorter carbon chains were more advantageous to emulsifying triglyceride compared with PCs having longer carbon chains [15]. Having long carbon chains lessened the dispersibility in water, while having too short a carbon chain impaired the emulsification ability. PCs having 6–10 carbons in their acyl hydrocarbon group were more advantageous to the formation of stable O/W emulsions because these PCs were still able to form bilayer structures and also had strong hydrophilic properties [15]. The reports supported the results of the current study, in which shorter chain PCs produced emulsions with smaller droplet sizes. The particle sizes in the O/W emulsion depended on the HLB that was influenced by the length of the hydrocarbon chains in the PC used as an emulsifier [10,11]. In contrast, the carbon numbers in acyl hydrocarbon chains in a TG, affected the particle sizes of the emulsion less, when using the same PC. The preparation temperature was selected at 65 ◦ C, which was sufficiently above the melting points of all TGs, an important factor in the preparation of lipid particles, particularly using lipids that are in solid form at room temperatures [7]. Under the condition that the preparation temperature was higher than the melting point of the TG to be emulsified, several potential influencing factors being different in each TG (e.g., physical size and viscosity) were considered not to be the dominant determining factors influencing the particle sizes of the emulsion. However, it was notable that longer chain TGs (trilaurin and trimyristin) showed a greatly increased mean diameter compared with the shorter chain TGs (tricaprylin and tricaprin). This tendency was more obvious when the TGs were emulsified by longer chain PCs (DPPC and DSPC), whereas the particle sizes were not largely varied when using shorter chain PCs (DLPC and DMPC). TGs having long acyl hydrocarbon chains were highly lipophilic and thus, more difficult to emulsify. Therefore, it was expected that the potency of the emulsifier directly affected the particle sizes of the emulsion droplets. The results of the multiple regression analysis using both the carbon numbers of the PC and TG as independent variables supported these findings. The mean diameter of the emulsion droplets could be estimated as a function of the carbon numbers of the PC and TG, and the carbon number of the PC was the dominant contributing factor. Egg yolk lecithin is a mixture of various phospholipids. It consists mostly of phosphatidylcholine, which accounts for
approximately 80% of the composition and the residual part includes phosphatidylethanolamine and several other phospholipids [15]. Since the carbon number of the acyl hydrocarbon chains in the PC was the factor that most significantly affected the particle sizes with the synthesized PC, we examined whether the composition of the acyl hydrocarbon chains in each lecithin also had an affect even though the polar head groups are heterogeneous in egg yolk lecithin. The arithmetic carbon number of each lecithin was calculated for analysis based on the composition of the acyl hydrocarbon chains described in Table 1, which resulted in 17.64 for PEL, 17.83 for R-20 and 17.86 for R-5. Since R-20 and R-5 were hydrogenated compounds of purified egg yolk lecithin and therefore, the major difference among these lecithins was the degree of saturation, the calculated carbon numbers of the three kinds of lecithins were all similar. Fig. 5 presents the experimental mean values of the particle sizes (bars) and the corresponding estimated values calculated with the multiple regression function derived from synthesized PC data using the carbon numbers mentioned above (dots and connected lines). In R-5, the estimated values were at a comparable level to the corresponding experimental mean values but the slope was steeper (14.5) than that of the experimental mean values (9.2). In R-20 and PEL, the estimated values were higher than the corresponding experimental mean values. As for PEL, the slope of the estimated values (14.5) was much steeper than that of the experimental mean values (5.7). The major differences between the synthesized PC and the egg yolk lecithin used in this study were: (1) all the egg yolk lecithin contained phosphatidylethanolamines (PE); (2) all the egg yolk lecithin contained a mixture of various length acyl hydrocarbon chains and (3) PEL and R-20 contained unsaturated hydrocarbon chains. The PEs and the unsaturated phospholipids both assumed the cone shape while the saturated PCs assumed the cylinder shape. Association of such cone shape molecules was weaker than that of cylinder shape molecules due to a longer intermolecular distance, which resulted in a higher fluidity of phospholipids containing PEs or unsaturated phospholipids [16]. The results for R-5, which was almost fully saturated, suggested that the carbon number of the lipophilic part of the lecithin was still a dominant factor determining the particle sizes of the emulsion droplets when using the lecithin consists of saturated acyl hydrocarbon chains. The reason for the gradual changes in particle sizes depending on TGs in R-5 compared with those of synthesized PCs were unclear. A possible explanation was that the influence of PEs probably weakened the association of the phospholipid molecules mentioned above and thus might have diminished the effects of the various TGs on the particle sizes in the emulsions. In addition, a variety of acyl hydrocarbon chain lengths might have allowed a flexible packing geometry of phospholipids at the interface. In R-20 and PEL, the influence of PEs, and moreover, unsaturated phospholipids, was expected to have a greater importance in determining the particle sizes of the emulsion droplets, which resulted in smaller particle sizes than the estimated
T. Nii, F. Ishii / Colloids and Surfaces B: Biointerfaces 41 (2005) 305–311
311
Fig. 5. Experimental data and the estimated value of the mean diameters of emulsion droplets using three kinds of egg yolk lecithins. Experimental data (bars) and corresponding estimated values calculated by the multiple regression function derived from synthesized PC data using the arithmetic carbon number for each lecithin (dots and connected lines).
value derived from the regression function, based on saturated PC data. Possible reasons as to why changes in particle sizes depending on TGs were gradual in PEL (5.7) but were comparable to synthesized PC in R-20 (18.4) remain to be elucidated.
5. Conclusion When using a synthesized saturated PC as an emulsifier, the carbon numbers of the PC had a strong correlation with the mean diameters of the emulsion droplets when analyzed with each TG. The carbon numbers of the TG had less of a correlation with the mean diameters than the PC in a simple regression analysis. Multiple regression analysis using the carbon numbers of both the PC and TG as independent variables showed a remarkable significance in the regression function and in all regression coefficients. Among the regression coefficients, the contribution of the carbon numbers of the PC was the most significant. These results indicated that a multiple regression function might be useful to estimate the mean diameters of the emulsion droplets in any combination of PC and TG used in this study. In the experiments using egg yolk lecithins, the mean diameters tended to increase in the order of PEL, R-20 and R-5 that correspond to the order of degrees of saturation in acyl hydrocarbon chains. Estimated values derived from the regression function based on saturated PC data were at a comparable level to the corresponding experimental mean values in R-5, while those in R-20 and PEL were higher than the experimental values. These findings gave useful information on the mean diameters of emulsion droplets when designing an emulsion formulation
using a particular combination of a phospholipid and triglyceride. References [1] K. Jores, W. Mehnert, M. Drechsler, H. Bunjes, C. Johann, K. M¨ader, J. Control. Release 95 (2004) 217–227. [2] R.H. M¨uller, S. Schmidt, I. Buttle, A. Akkar, J. Schmitt, S. Br¨omer, Int. J. Pharm. 269 (2004) 293–302. [3] M. Jumaa, B.W. M¨uller, Int. J. Pharm. 163 (1998) 81–89. [4] R.P. Bagwe, J.R. Kanicky, B.J. Palla, P.K. Patanjali, D.O. Shah, Crit. Rev. Ther. Drug Carrier Syst. 18 (2001) 77–140. [5] S. Benita, M.Y. Levy, J. Pharm. Sci. 82 (1993) 1069–1079. [6] M. Trotta, F. Pattarino, T. Ignoni, Eur. J. Pharm. Biopharm. 53 (2002) 203–208. [7] B. Heurtault, P. Saulnier, B. Pech, J. Proust, J. Benoit, Biomaterials 24 (2003) 4283–4300. [8] H. Heiati, N.C. Phillips, R. Tawashi, Pharm. Res. 13 (1996) 1406–1410. [9] B. Heurtault, P. Saulnier, B. Pech, M.C. Venier-Julienne, J.E. Proust, R. Phan-Tan-Luu, J.P. Benoit, Eur. J. Pharm. Sci. 18 (2003) 55–61. [10] T. Nii, F. Ishii, Colloids Surf. B Biointerfaces 39 (2004) 57–63. [11] F. Ishii, T. Nii, Colloids Surf. B Biointerfaces 41 (2005) 257–262. [12] Asahi Chemicals Industries Co. Ltd, in house analysis data, unpublished. [13] F. Ishii, I. Sasaki, H. Ogata, J. Pharm. Pharmacol. 42 (1990) 513–515. [14] S.S. Davis, J. Hadgraft, K.J. Palin, in: P. Becher (Ed.), Encyclopedia of Emulsion Technology: Applications, Marcel Dekker, New York, 1987, pp. 165–171. [15] S.S. Davis, in: I.D.A. Johnston (Ed.), Current Perspectives in the Use of Lipid Emulsion, MTP Press, Lancaster, England, 1982, pp. 35–61. [16] P.R. Cullis, M.J. Hope, B. de Kruijff, A.J. Verkleij, C.P.S. Tilcock, in: J.F. Kuo (Ed.), Phospholipid and Cellular Regulation, vol. 1, CRC Press, Boca Raton, Florida, 1985, pp. 1–59.
Dialkylphosphatidylcholine and egg yolk lecithin for emulsification of various triglycerides Tomoko Nii, Fumiyoshi Ishii ∗ Department of Pharmaceutical Sciences, Meiji Pharmaceutical University, 2-522-1 Noshio, Kiyose, Tokyo 204-8588, Japan Received 27 October 2004; accepted 23 December 2004
Abstract Synthesized saturated phosphatidylcholine (PC) and egg yolk lecithin (EYL) were investigated to explore their influence on particle sizes in emulsions when dispersing various triglycerides (TG). One of four different kinds of synthesized saturated PC (DLPC, DMPC, DPPC and DSPC) or three different kinds of EYL (purified EYL (PEL) and hydrogenated purified EYL with two different iodine values (IV), R-20 and R-5), 2.5% (w/w) glycerol solution and one of four kinds of TG (tricaprylin, tricaprin, trilaurin and trimyristin) were sonicated five times for 1 min with intervals of 0.5 min. When using four kinds of synthesized saturated PCs as emulsifiers, the carbon numbers of each PC had a strong correlation with the mean diameters of the emulsion when analyzed with each of the four kinds of TG used in the study (regression function ranged from 0.811 to 0.915). The carbon numbers of the TG had less correlation with the mean diameters than the PC in simple regression analysis (regression function ranged from 0.236 to 0.875). Multiple regression analysis using the carbon numbers both of the PC and TG as independent variables was remarkably significant in the regression function (2.0 × 10−14 ) and all regression coefficients (2.7 × 10−13 , 5.8 × 10−7 and 1.9 × 10−9 for PC, TG and intercept, respectively). Among the regression coefficients, the contribution of the carbon number of the PC was the most significant. These results indicated that a multiple regression function should be useful to estimate the mean diameters of emulsion droplets in any combinations of PC and TG used in this study. In the experiments using three kinds of EYL, the mean diameters also tended to increase according to the order of PEL, R-20 and R-5, which corresponds to the order of degrees of saturation (IV = 75, 20 and 2, respectively). The experimental values for EYL were compared with the estimated values calculated by the multiple regression function derived from synthesized PC data using the arithmetic carbon number, based on the components of each EYL. The estimated mean diameters were at comparable levels to the corresponding experimental mean diameters in the most saturated hydrogenated lecithin (R-5), while those were larger than the experimental mean diameters in two less saturated kinds of lecithin (R-20 and purified EYL). These findings gave useful information on the mean diameters of emulsion droplets when designing an emulsion formulation using a particular combination of a phospholipid and triglyceride. © 2005 Elsevier B.V. All rights reserved. Keywords: Phosphatidylcholine; Purified egg yolk lecithin; Hydrogenated egg yolk lecithin; Emulsion; Triglyceride; Particle size; Hydrophilic–lipophilic balance
1. Introduction Submicron oil-in-water (O/W) emulsions are expected to be a practical and promising option to disperse drugs that have the problem of poor water solubility into an aqueous medium [1,2]. There are numerous variations in the emulsifier system, but the surfactants and oil phases commonly ∗
Corresponding author. Tel.: +81 424 95 8468; fax: +81 424 95 8468. E-mail address: [email protected] (F. Ishii).
0927-7765/$ – see front matter © 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.colsurfb.2004.12.017
used for parenteral administration are limited [3]. Lecithins are the most common surfactant [4–6], while vegetable oils or medium-chain triglycerides (MCT) are commonly used for the oil phase in parenteral emulsions [3]. In recent years, melt-emulsified nanoparticles based on lipids that are solid at room temperature have been more prevalent [1,7,8]. Such solid lipid nanoparticles (SLN) prepared with triglycerides (TGs) and phospholipids have the advantages of a longer retention of the incorporated drugs than lipid emulsions and a more rapid degradation in the body than polymeric
306
T. Nii, F. Ishii / Colloids and Surfaces B: Biointerfaces 41 (2005) 305–311
nanoparticles prepared with polymers [8]. The SLN bears many similarities to the lipid emulsion and also liposomes, for phospholipids are on the surface of the particulates in these colloidal systems [8]. For intravenous application, particle size is considered a major issue since the half-lives of SLN were found to be dependent on their size [9]. In addition, variations of the particle size within the colloidal range may affect the physicochemical properties of the nanoparticles [9]. Accordingly, the particle size of the colloidal lipid particles is an important characteristic and thus, potential factors influencing the particle size are of interest. We previously studied the characteristics of various saturated and unsaturated phosphatidylcholines (PCs) used for emulsifying a MCT [10,11]. Particle sizes were influenced by the length and degrees of unsaturation of the acyl hydrocarbon chains of the PC. We suggested that the physical length and/or the shape of hydrocarbon chains of the PC and hydrophilic–lipophilic balance (HLB) of the PC might influence the particle sizes of the emulsion [10]. The primary objective of the current study was to investigate the characteristics of synthesized saturated PCs that possibly influence the particle sizes of the emulsion using various triglycerides. The relationship between the hydrocarbon chains in the PC used as the emulsifier and the hydrocarbon chains in the TG as the oil phase is a matter of particular interest. In addition, purified egg yolk lecithin and its hydrogenated compounds were used as mixtures of various phospholipids to explore the differences from pure PCs when emulsifying various TGs.
2. Materials and methods 2.1. Materials The following synthesized saturated PCs were purchased from NOF Corporation (Tokyo, Japan): l-␣-phosphatidylcholine dilauroyl (DLPC), l-␣-phosphatidylcholine dimyristoyl (DMPC), l-␣-phosphatidylcholine dipalmitoyl (DPPC) and l-␣-phosphatidylcholine distearoyl (DSPC). Glycerol trioctanoate (tricaprylin), glycerol tridecanoate (tricaprin), glyceryl trilaurate (trilaurin) and glycerol trimyristate (trimyristin) were purchased from Tokyo Kasei Kogyo Co. Ltd. (Tokyo, Japan). Purified egg yolk lecithin (PEL, iodine value, IV = 75), hydrogenated purified egg yolk lecithin with two different iodine values, i.e., R-20 (IV = 20) and R-5 (IV = 2) were kindly provided by Asahi Chemical Industries Co. Ltd. (Tokyo, Japan). The compositions of the acyl hydrocarbon chains for these lecithins are shown in Table 1 [12]. The glycerol was of the highest grade, which was a product of Wako Pure Chemical Industries (Osaka, Japan). 2.2. Preparation of the emulsion The emulsion was prepared by probe sonication. Each component of the emulsion was weighed into a glass container according to the ratio, as previously reported
Table 1 Composition of acyl hydrocarbon chains in egg yolk lecithin [12] PEL
R-20
R-5
C-16 16:0 16:1
30.4 1.5
29.4 0.0
29.3 0.0
C-18 18:0 18:1 18:2 18:3
15.2 27.7 15.5 0.0
35.8 20.0 0.0 0.0
55.8 0.0 0.0 0.0
C-20 20:0 20:1 20:2 20:4
0.0 0.0 0.0 5.5
3.2 3.6 0.6 0.0
7.7 0.0 0.0 0.0
C-22 22:0 22:1 22:2 22:5 22:6
0.0 0.0 0.0 0.0 4.1
2.9 3.0 1.2 0.0 0.0
7.0 0.0 0.0 0.0 0.0
Others
0.1
0.2
0.2
Data represent relative amounts (%).
[10,11,13], i.e., one of four kinds of synthesized PC or three kinds of lecithin (120 mg), 2.5% (w/w) glycerol solution (9.0 g) and one of four kinds of triglyceride (1.0 g). The mixture was sonicated five times for 1 min with intervals of 0.5 min with a probe ultrasonicator (VP-30S, Taitec Corporation, Japan, 20 kHz). The sonication was conducted at 65 ◦ C, which was sufficiently above the phase transition temperatures of all phospholipids, as well as the melting points of all triglycerides used in this study. 2.3. Particle size measurement Mean diameters of droplets in O/W emulsions were measured by a submicron particle size analyzer (NICOMP, Model 370, Pacific Scientific Instrument Division Co. Ltd.). 2.4. Statistics A statistical analysis was performed to examine the correlation between the mean diameters and carbon numbers of the acyl hydrocarbon chains in synthesized PCs or TGs. The following numbers were used for analysis. Synthesized PC: DLPC; 12, DMPC; 14, DPPC; 16 and DSPC; 18. TG: tricaprylin; 8, tricaprin; 10, trilaurin; 12 and trimyristin; 14. The correlation between the carbon numbers and the mean diameters was analyzed using Spearman’s correlation coefficient with the rank test. Moreover, a regression analysis was performed to examine if the carbon numbers of the acyl
T. Nii, F. Ishii / Colloids and Surfaces B: Biointerfaces 41 (2005) 305–311
hydrocarbon chains in synthesized PCs or TGs were associated with the mean diameters.
3. Results 3.1. Emulsion using synthesized saturated PC The results for the emulsion using synthesized PCs are presented in Figs. 1 and 2. Fig. 1 has the data grouped according to the TGs used for the oil phase. With all TGs, the mean diameters of the emulsion droplets clearly increased as the number of carbons in the acyl hydrocarbon chains in the PC increased. The same data grouped according to the PCs used as the emulsifier are shown in Fig. 2. The mean diameters also tended to increase as the number of carbons
Fig. 1. Mean diameters of emulsion droplets using four kinds of TGs with different hydrocarbon chain lengths (mean ± S.D., n = 3). Each TG was emulsified with four kinds of synthesized PCs with different hydrocarbon chain lengths.
Fig. 2. Mean diameters of emulsion droplets using four kinds of synthesized PCs with different hydrocarbon chain lengths (mean ± S.D., n = 3). Each PC was used to emulsify four kinds of TGs with different hydrocarbon chain lengths.
307
in the acyl hydrocarbon chains of TG increased. The difference in the particle sizes among the TGs was less clear in the shorter chain PCs (DLPC, DMPC), e.g., the mean diameters in tricaprylin, tricaprin, trilaurin and trimyristin were 116.4, 124.0, 144.3 and 147.0 in DLPC, respectively. In contrast, the particle sizes with the longer chain PCs (DPPC, DSPC) were more dependent on the number of carbons in the acyl hydrocarbon chains, i.e., the mean diameters of the above four TGs in DSPC were 204.4, 211.9, 335.5 and 390.8, respectively. With all PCs, to greater or lesser degrees, the particle sizes prepared with tricaprylin and tricaprin were similar while those prepared with trilaurin and trimyristin tended to increase more than these two shorter chain TGs. To ascertain whether the carbon numbers of the acyl hydrocarbon chains in the synthesized PCs or TGs were potential factors affecting the particle sizes of emulsion droplets, the correlation of the mean diameters and carbon numbers was statistically analyzed. Table 2 summarizes the coefficients and p-values of these analyses. The correlation coefficients and p-values obtained from Spearman’s correlation coefficient with the rank test are shown in the table. With regard to the regression analysis, the coefficient of determination and the p-values of the regression function tested by ANOVA were demonstrated to examine if the regression function was useful to estimate the mean diameters when using carbon numbers as explanatory variables. The significance of the regression coefficients, i.e., the intercept and independent variable(s), were also presented to examine if these factors were necessary to estimate the dependent variable. First, a simple regression analysis was performed for each TG (Group A, corresponding to Fig. 1) and each PC (Group B, corresponding to Fig. 2). In Group A, the carbon numbers of PC had a strong correlation with the mean diameter of emulsion when analyzed with each TG. The correlation coefficient (rs) in each TG ranged from 0.907 to 0.972, which meant that the mean diameters had a very strong correlation with the carbon number of the PC in every oil phase. Simple regression analysis also supported this strong correlation. The coefficient of determination in each TG was high, ranging from 0.811 to 0.915. The p-values of the regression lines were all below 0.0001, which was highly significant. These results demonstrated that the regression line of the carbon numbers of the PC was useful to estimate the mean diameter of the emulsion droplets. The regression coefficient was also significant in both its intercept and independent variable. However, the intercept was much less significant than the independent variable, which meant that the independent variables contributed more to the regression line. In Group B, the carbon numbers of the TG showed less of a correlation with the mean diameter than the PC. The correlation coefficient ranged from 0.561 to 0.907 and the coefficient of determination from 0.236 to 0.875. DMPC was not significant in Spearman’s correlation coefficient according to the rank test and regression analysis. It was notable that longer chain PCs
308
T. Nii, F. Ishii / Colloids and Surfaces B: Biointerfaces 41 (2005) 305–311
Table 2 Statistical analysis of correlation coefficient and regression analysis among particle sizes, carbon numbers of the phosphatidylcholine and carbon numbers of the triglyceride in emulsion using synthesized phosphatidylcholine Group
Independent variable 1
Independent variable 2
Synthesized phosphatidylcholine Group A: Each TG C number of PC (12–18) TG-C8 TG-C10 TG-C12 TG-C14
n
Correlation coefficienta
Regression analysisb
Correlation p coefficient
Regression function
Regression coefficient
Coefficient of determination
p
Intercept, p
Independent Independent variable 1, p variable 2, p
12 12 12 12
0.972 0.972 0.907 0.907
0.0013 0.0013 0.0026 0.0026
0.870 0.915 0.832 0.811
0.000006 0.000001 0.000022 0.000040
0.0260 0.0102 0.0026 0.0018
0.000006 0.000001 0.000022 0.000040
Group B: Each PC C number of TG (8–14) PC-C12 PC-C14 PC-C16 PC-C18
12 12 12 12
0.907 0.561 0.842 0.907
0.0026 0.0626 0.0052 0.0026
0.501 0.236 0.787 0.875
0.006012 0.062259 0.000073 0.000005
0.0028 0.0178 0.0212 0.0650
0.006012 0.062259 0.000073 0.000005
Group C: All sample Simple regression analysis C number of PC (12–18) C number of TG (8–14)
48 0.831 48 0.433
1.2 × 10−8 0.560 0.0030 0.167
5.9 × 10−10 0.002287
0.0004 0.5348
5.9 × 10−10 0.002287
0.743
2.0 × 10−14
1.9 × 10−9
2.7 × 10−13
Multiple regression analysis C number of PC (12–18)
C number of TG (8–14)
48
5.8 × 10−7
Abbreviations: TG-C8, tricaprylin; TG-C10, tricaprin; TG-C12, trilaurin; TG-C14, trimyristin; PC-C12, DLPC; PC-C14, DMPC; PC-C16, DPPC; PC-C18, DSPC. a Spearman’s correlation coefficient by rank test was used to examine correlation between the C numbers and mean diameters. b Groups A and B were examined by simple regression analysis while Group C was examined by both simple regression analysis and multiple regression analysis using the C numbers of the TG and PC of each sample.
(DPPC and DSPC) tended to indicate a stronger correlation with the carbon numbers of the TG than the shorter chain PCs. Next, all data were examined by simple regression and multiple regression analysis (Group C). In simple regression analysis, all data were tested using only a single independent variable, i.e., the carbon numbers of the PC or TG. A strong correlation was still maintained between the carbon numbers of the PC and the mean diameters, regardless of the TG in the oil phase, but the correlation coefficient and in particular the coefficient of determination were lowered. It was anticipated from the results of Group B that a correlation between the carbon numbers of the TG and the mean diameters was considerably lower, even though it was still significant. The correlation coefficient was 0.433 and the coefficient of determination was as low as 0.167, mainly due to the highly insignificant level of the intercept in the regression line. Multiple regression analysis was then performed using both the carbon numbers of the PC (variable 1) and TG (variable 2) as independent variables. The results of the analysis were remarkably significant, that is, the coefficient of determination was 0.743, for which the p-value was 2.0 × 10−14 . Concerning the regression coefficient, the p-values of variable 1, variable 2 and the intercept were 2.7 × 10−13 , 5.8 × 10−7 and 1.9 × 10−9 ,
respectively, and the contribution of the carbon numbers of the PC (variable 1) was most significant among these three regression coefficients. These results indicated that a multiple regression function might be useful to estimate the mean diameters of emulsion droplets in any combination of PC and TG used in this study. 3.2. Emulsion using egg yolk lecithin Figs. 3 and 4present the results of emulsion using egg yolk lecithin. Fig. 3 grouped the data by TG used for the oil phase. The mean diameters also tended to increase according to the order of PEL, R-20 and R-5, which corresponds to the order of degrees of saturation (IV = 75, 20 and 2, respectively). R20 fell between PEL and R-5, and the mean diameters of R-20 were closer to PEL in the shorter chain TGs while they were closer to R-5 in the longer chain TGs. Fig. 4 shows the same data grouped by the lecithin used as the emulsifier. For all lecithins, the mean diameters of the emulsion droplets increased as the number of carbons in the acyl hydrocarbon chains in TG increased. For egg yolk lecithin, the correlation between the carbon numbers of the TG and the mean diameters was statistically analyzed, and the results are summarized in Table 3. Simple
T. Nii, F. Ishii / Colloids and Surfaces B: Biointerfaces 41 (2005) 305–311
Fig. 3. Mean diameters of emulsion droplets using four kinds of TGs with different hydrocarbon chain lengths (mean ± S.D., n = 3). Each TG was emulsified with three kinds of egg yolk lecithins with different degrees of hydration.
regression analysis was performed for each lecithin (Group D, corresponding to Fig. 4). The carbon numbers of the TG indicated a significant correlation, but the degrees of significance were not as strong in PEL and R-5. The correlation coefficient was 0.756 and 0.713, and the coefficient of determination was 0.441 and 0.375 in PEL and R-5, respectively. For these lecithins, the contribution of the independent variable, i.e., the carbon numbers of the TG, to the regression line was rather weak. In contrast, R-20 showed a stronger correlation with the carbon numbers of the TG. The regression line was also highly significant, though the intercept contributed least to the regression line. Then, all data were examined by simple regression analysis. A significant correlation between
309
Fig. 4. Mean diameters of emulsion droplets using three kinds of egg yolk lecithins with different degrees of hydration (mean ± S.D., n = 3). Each egg yolk lecithin was used to emulsify four kinds of TGs with different hydrocarbon chain lengths.
the carbon numbers of the TG and the mean diameters was still evident. The correlation coefficient was 0.580 and the coefficient of determination was 0.313.
4. Discussion Particle sizes of O/W emulsions were influenced by various factors, for instance, dispersed oil and its ratio to the continuous water phase, the emulsifier and its concentration in total colloidal solution and the method of emulsion preparation [14]. In this study, various combinations of TGs and phospholipids were dispersed in glycerol solution by probe
Table 3 Statistical analysis of the correlation coefficient and regression analysis between particle sizes and carbon numbers of the triglyceride in emulsion using egg yolk lecithin Group
Independent variable 1
Independent n variable 2
Correlation coefficienta
Regression analysisb
Correlation p coefficient
Regression function
Regression coefficient
Coefficient of p determination
Intercept, p Independent variable 1, p
Egg yolk lecithin Group D: Each EYL PEL C number of TG (8–14) R-20 R-5
12 0.756 12 0.928 12 0.713
0.0122 0.0021 0.0181
0.441 0.807 0.375
0.011036 0.0001 0.000044 0.2958 0.020188 0.0015
0.011036 0.000044 0.020188
Group E: All sample Simple regression analysis C number of TG (8–14)
36 0.580
0.0006
0.313
0.000229 0.0009
0.000229
Independent variable 2, p
Abbreviations: EYL, egg yolk lecithin; PEL, purified egg yolk lecithin (IV = 75); R-20, hydrogenated purified egg yolk lecithin (IV = 20); R-5, hydrogenated purified egg yolk lecithin (IV = 2). a Spearman’s correlation coefficient by rank test was used to examine the correlation between C numbers and mean diameters. b Simple regression analysis was used in both Groups D and E.
310
T. Nii, F. Ishii / Colloids and Surfaces B: Biointerfaces 41 (2005) 305–311
sonication to investigate how their characteristics affected the particle sizes of the emulsion droplets. Concerning the synthesized PCs, the particle sizes of the emulsion droplets showed a strong correlation with the carbon numbers of the acyl hydrocarbon chains of PC when dispersing the same TG (rs = 0.907–0.972). In a previous study using tricaprylin, the mean diameters showed a strongly positive correlation with the carbon number of the PC when the emulsion was prepared at 65 ◦ C, which was sufficiently above the transition temperatures of all PCs used [10]. The results were consistent with our current study not only in tricaprylin but also other TGs having longer acyl hydrocarbon chains. It was reported that PCs with shorter carbon chains were more advantageous to emulsifying triglyceride compared with PCs having longer carbon chains [15]. Having long carbon chains lessened the dispersibility in water, while having too short a carbon chain impaired the emulsification ability. PCs having 6–10 carbons in their acyl hydrocarbon group were more advantageous to the formation of stable O/W emulsions because these PCs were still able to form bilayer structures and also had strong hydrophilic properties [15]. The reports supported the results of the current study, in which shorter chain PCs produced emulsions with smaller droplet sizes. The particle sizes in the O/W emulsion depended on the HLB that was influenced by the length of the hydrocarbon chains in the PC used as an emulsifier [10,11]. In contrast, the carbon numbers in acyl hydrocarbon chains in a TG, affected the particle sizes of the emulsion less, when using the same PC. The preparation temperature was selected at 65 ◦ C, which was sufficiently above the melting points of all TGs, an important factor in the preparation of lipid particles, particularly using lipids that are in solid form at room temperatures [7]. Under the condition that the preparation temperature was higher than the melting point of the TG to be emulsified, several potential influencing factors being different in each TG (e.g., physical size and viscosity) were considered not to be the dominant determining factors influencing the particle sizes of the emulsion. However, it was notable that longer chain TGs (trilaurin and trimyristin) showed a greatly increased mean diameter compared with the shorter chain TGs (tricaprylin and tricaprin). This tendency was more obvious when the TGs were emulsified by longer chain PCs (DPPC and DSPC), whereas the particle sizes were not largely varied when using shorter chain PCs (DLPC and DMPC). TGs having long acyl hydrocarbon chains were highly lipophilic and thus, more difficult to emulsify. Therefore, it was expected that the potency of the emulsifier directly affected the particle sizes of the emulsion droplets. The results of the multiple regression analysis using both the carbon numbers of the PC and TG as independent variables supported these findings. The mean diameter of the emulsion droplets could be estimated as a function of the carbon numbers of the PC and TG, and the carbon number of the PC was the dominant contributing factor. Egg yolk lecithin is a mixture of various phospholipids. It consists mostly of phosphatidylcholine, which accounts for
approximately 80% of the composition and the residual part includes phosphatidylethanolamine and several other phospholipids [15]. Since the carbon number of the acyl hydrocarbon chains in the PC was the factor that most significantly affected the particle sizes with the synthesized PC, we examined whether the composition of the acyl hydrocarbon chains in each lecithin also had an affect even though the polar head groups are heterogeneous in egg yolk lecithin. The arithmetic carbon number of each lecithin was calculated for analysis based on the composition of the acyl hydrocarbon chains described in Table 1, which resulted in 17.64 for PEL, 17.83 for R-20 and 17.86 for R-5. Since R-20 and R-5 were hydrogenated compounds of purified egg yolk lecithin and therefore, the major difference among these lecithins was the degree of saturation, the calculated carbon numbers of the three kinds of lecithins were all similar. Fig. 5 presents the experimental mean values of the particle sizes (bars) and the corresponding estimated values calculated with the multiple regression function derived from synthesized PC data using the carbon numbers mentioned above (dots and connected lines). In R-5, the estimated values were at a comparable level to the corresponding experimental mean values but the slope was steeper (14.5) than that of the experimental mean values (9.2). In R-20 and PEL, the estimated values were higher than the corresponding experimental mean values. As for PEL, the slope of the estimated values (14.5) was much steeper than that of the experimental mean values (5.7). The major differences between the synthesized PC and the egg yolk lecithin used in this study were: (1) all the egg yolk lecithin contained phosphatidylethanolamines (PE); (2) all the egg yolk lecithin contained a mixture of various length acyl hydrocarbon chains and (3) PEL and R-20 contained unsaturated hydrocarbon chains. The PEs and the unsaturated phospholipids both assumed the cone shape while the saturated PCs assumed the cylinder shape. Association of such cone shape molecules was weaker than that of cylinder shape molecules due to a longer intermolecular distance, which resulted in a higher fluidity of phospholipids containing PEs or unsaturated phospholipids [16]. The results for R-5, which was almost fully saturated, suggested that the carbon number of the lipophilic part of the lecithin was still a dominant factor determining the particle sizes of the emulsion droplets when using the lecithin consists of saturated acyl hydrocarbon chains. The reason for the gradual changes in particle sizes depending on TGs in R-5 compared with those of synthesized PCs were unclear. A possible explanation was that the influence of PEs probably weakened the association of the phospholipid molecules mentioned above and thus might have diminished the effects of the various TGs on the particle sizes in the emulsions. In addition, a variety of acyl hydrocarbon chain lengths might have allowed a flexible packing geometry of phospholipids at the interface. In R-20 and PEL, the influence of PEs, and moreover, unsaturated phospholipids, was expected to have a greater importance in determining the particle sizes of the emulsion droplets, which resulted in smaller particle sizes than the estimated
T. Nii, F. Ishii / Colloids and Surfaces B: Biointerfaces 41 (2005) 305–311
311
Fig. 5. Experimental data and the estimated value of the mean diameters of emulsion droplets using three kinds of egg yolk lecithins. Experimental data (bars) and corresponding estimated values calculated by the multiple regression function derived from synthesized PC data using the arithmetic carbon number for each lecithin (dots and connected lines).
value derived from the regression function, based on saturated PC data. Possible reasons as to why changes in particle sizes depending on TGs were gradual in PEL (5.7) but were comparable to synthesized PC in R-20 (18.4) remain to be elucidated.
5. Conclusion When using a synthesized saturated PC as an emulsifier, the carbon numbers of the PC had a strong correlation with the mean diameters of the emulsion droplets when analyzed with each TG. The carbon numbers of the TG had less of a correlation with the mean diameters than the PC in a simple regression analysis. Multiple regression analysis using the carbon numbers of both the PC and TG as independent variables showed a remarkable significance in the regression function and in all regression coefficients. Among the regression coefficients, the contribution of the carbon numbers of the PC was the most significant. These results indicated that a multiple regression function might be useful to estimate the mean diameters of the emulsion droplets in any combination of PC and TG used in this study. In the experiments using egg yolk lecithins, the mean diameters tended to increase in the order of PEL, R-20 and R-5 that correspond to the order of degrees of saturation in acyl hydrocarbon chains. Estimated values derived from the regression function based on saturated PC data were at a comparable level to the corresponding experimental mean values in R-5, while those in R-20 and PEL were higher than the experimental values. These findings gave useful information on the mean diameters of emulsion droplets when designing an emulsion formulation
using a particular combination of a phospholipid and triglyceride. References [1] K. Jores, W. Mehnert, M. Drechsler, H. Bunjes, C. Johann, K. M¨ader, J. Control. Release 95 (2004) 217–227. [2] R.H. M¨uller, S. Schmidt, I. Buttle, A. Akkar, J. Schmitt, S. Br¨omer, Int. J. Pharm. 269 (2004) 293–302. [3] M. Jumaa, B.W. M¨uller, Int. J. Pharm. 163 (1998) 81–89. [4] R.P. Bagwe, J.R. Kanicky, B.J. Palla, P.K. Patanjali, D.O. Shah, Crit. Rev. Ther. Drug Carrier Syst. 18 (2001) 77–140. [5] S. Benita, M.Y. Levy, J. Pharm. Sci. 82 (1993) 1069–1079. [6] M. Trotta, F. Pattarino, T. Ignoni, Eur. J. Pharm. Biopharm. 53 (2002) 203–208. [7] B. Heurtault, P. Saulnier, B. Pech, J. Proust, J. Benoit, Biomaterials 24 (2003) 4283–4300. [8] H. Heiati, N.C. Phillips, R. Tawashi, Pharm. Res. 13 (1996) 1406–1410. [9] B. Heurtault, P. Saulnier, B. Pech, M.C. Venier-Julienne, J.E. Proust, R. Phan-Tan-Luu, J.P. Benoit, Eur. J. Pharm. Sci. 18 (2003) 55–61. [10] T. Nii, F. Ishii, Colloids Surf. B Biointerfaces 39 (2004) 57–63. [11] F. Ishii, T. Nii, Colloids Surf. B Biointerfaces 41 (2005) 257–262. [12] Asahi Chemicals Industries Co. Ltd, in house analysis data, unpublished. [13] F. Ishii, I. Sasaki, H. Ogata, J. Pharm. Pharmacol. 42 (1990) 513–515. [14] S.S. Davis, J. Hadgraft, K.J. Palin, in: P. Becher (Ed.), Encyclopedia of Emulsion Technology: Applications, Marcel Dekker, New York, 1987, pp. 165–171. [15] S.S. Davis, in: I.D.A. Johnston (Ed.), Current Perspectives in the Use of Lipid Emulsion, MTP Press, Lancaster, England, 1982, pp. 35–61. [16] P.R. Cullis, M.J. Hope, B. de Kruijff, A.J. Verkleij, C.P.S. Tilcock, in: J.F. Kuo (Ed.), Phospholipid and Cellular Regulation, vol. 1, CRC Press, Boca Raton, Florida, 1985, pp. 1–59.