MASTER’S PROGRAMME IN PHARMACEUTICAL MODELLING HT-20
MASTER THESIS 45 HP
Effect of lipid-based formulation on the solubilization patterns of poorly water- soluble drugs.
Author:
Manjiri Subodh Gude
Supervisors:
Aleksei Kabedev, Albin Parrow, Per Larsson Department of Pharmacy, Uppsala University
26th May 2021
Abstract 3
1. Introduction 3
1.1. Purpose 5
2. Materials and Methods 5
2.1. Experimental section 5
2.1.1. Chemicals 5
2.1.4. LBF type IIIA-LC preparation 6
2.1.5. Drug solubility in LBF dispersed in water 6
2.1.6. HPLC-UV analysis 6
2.2. Computational section 7
2.2.1. Simulation systems of LBF 7
2.2.2. Simulation process 8
3. Results 9
3.1. Experimental section 9
3.1.1. Solubility studies of LBF 9
3.1.2. pH characterization of DIF 9
3.2. Computational section 10
3.2.1. LBF characterization and structure 10
3.2.2. Drug loading 11
4. Discussion 15
5. Conclusion 16
6. References 16
7. Appendix 19
Abstract
Poorly water-soluble drugs (PWSDs), to date, require advanced formulation techniques to improve solubility and achieve the required plasma concentration to show a therapeutic effect when orally administered. Lipid-based formulations (LBFs) are an enabling strategy that is being used to improve the oral delivery of PWSDs. The aim of this study was to investigate the effect of lipid-based formulation, Type IIIA-LC, on the solubilization patterns of PWSDs, namely, carvedilol and felodipine. Solubility studies, for both drugs, were performed with LBF dispersed in -1) dog intestinal fluid (DIF), and 2) water, to identify and compare the extent of solubility in different matrices, and in silico to identify interesting patterns with any correlations in experimental and computational data. Solubility studies showed that carvedilol had better solubility in LBF when compared to felodipine. Computational studies showed that both drugs solubilized in the colloid in both digested and undigested states. Effect of drug loading had no significant difference on the solubilization patterns of both drugs. The maximum drug loading done was for 100 molecules though there is the possibility of the colloid having a higher capacity. Digestion did not seem to have a significant effect on the distribution of both drugs. In vitro and in silico data were in qualitative agreement and therefore, this computational model can be further used to study the specific processes causing solubilization, improvement, and development of new LBFs.
Keywords: Lipid based formulations (LBFs), poorly water-soluble drugs (PWSDs), solubility studies, coarse grained molecular dynamic simulations, solubilization, dog intestinal fluid (DIF).
1. Introduction
Numerous emerging trends in combinatorial chemistry and drug design have led to drug candidates with increased lipophilicity and poor water solubility. This has led to suboptimal drug delivery and thereby decreased bioavailability. Specifically, orally delivered drugs face such problems. Recent studies have revealed that discovery and development of new drugs alone are often not sufficient to achieve therapeutic excellence and capture market economies.
Therefore, modified formulations of existing drugs are gaining more importance (Kalepu and Nekkanti, 2015). Carvedilol and felodipine will be used in this study. Carvedilol is a weak base that is substantially insoluble in water, acidic solutions, and gastric and intestinal fluids;
it is classified as a Class II drug in the Biopharmaceutical Classification System. The solubility of carvedilol varies according to the solvent pH (Alves, Prado and Rocha, 2016),whereas felodipine is a neutral compound and its physicochemical properties are yet unresolved. To overcome poor solubility and dissolution, several strategies have been
experimented, namely – emulsions, liposomes, nanosuspension, solid-dispersion, solid lipid nanoparticles and others (Rodriguez-Aller et al., 2015). Trends for “enabling formulations”- formulations that improve drugs bioavailability, are fostering interest amongst researchers and pharma companies. Development of enabling formulations is currently being guided by the following (simplified) hypothesis: if a poorly soluble drug (BCS class II drug) can be transferred into a solubilized state, one can achieve an absorption profile close to that of a soluble drug (BCS class I drug)(Buckley et al., 2013). One such enabling formulation strategy that has shown improvement over the years are the lipid-based formulation (LBFs).
LBF is a mixed micellar system that contains lipid digestion products like oils, surfactants, and organic solvents in varying ratios which makes it very complex. For PWSDs, the major advantages associated with LBFs are avoidance of rate-limiting dissolution, and improved intestinal solubility, caused by increased solubilization when the formulation components mix with the GI fluids (Larsson, Alskär and Bergström, 2017; Alskär et al., 2019). Some of the major disadvantages linked to LBFs are their complexity, chemical instability of pre-dissolved drugs and a limited understanding of the influence of LBF intestinal digestion on drug absorption (Alskär et al., 2019).To optimize the choice of lipid formulation for a particular drug, it is beneficial to understand the mechanism of LBF, significance of digestibility, solvent capacity of LBF to solubilize the drug dose, effect of excipient selection on LBF and self-emulsifying ability of LBF upon dilution. In the current study, we shed light upon the effect of type IIIA-LC LBF and how it affected the solubilization of PWSDs.
The Lipid Formulation Classification System (LFCS) divides lipid formulations as such: Type I, II, III and IV formulations (Pouton and Porter, 2008). Many of the marketed products are Type III systems but this group is particularly diverse because of the wide variation in the proportions of oily and water-soluble materials used. This group has been further divided into Type IIIA and Type IIIB, to distinguish between formulations which contain a significant proportion of oils (Type IIIA) and those which are predominantly water-soluble (Type IIIB) (Buckley et al., 2013). Type IIIA formulations usually contain 40–80% oils and 20–40%
water-soluble surfactants as well as 0–40% hydrophilic co-solvents (Xiao et al., 2016).
Type IIIA-LC formulations have the potential to disperse quickly to form fine submicron dispersions, often fine enough to form transparent dispersions. The key to successful formulation of Type III systems is to avoid formulations that are so hydrophilic that they lose a considerable proportion of their solvent capacity on dispersion. One of the advantages of type IIIA-LC LBF is that they facilitate drug absorption without the need for digestion (Pouton and Porter, 2008).
Dog intestinal fluid (DIF) was used to explore the solubilization effect of the system, formed by, carvedilol and felodipine, and the LBF- type IIIA-LC from Alskär et al’s study. As dogs are a common study animal for drug testing, a computational model of DIF treated with LBF could reduce the number of animal experiments and shed light on the LBF solubilization patterns. This will help investigate how PWSDs are solubilized in lipid-based formulations in DIF, as compared to LBF dispersed in water. A theoretical model mimicking the DIF with type IIIA-LC LBF and drug, will allow us to study the phase behavior of the system as well as the dissolution of the PWSDs.
1.1. Purpose
The study aimed to use MD simulations to understand molecular mechanisms behind the results observed experimentally using high-performance liquid chromatography (HPLC). The correlations between the computational and experimental datasets would help in gaining more insight on the apparent solubility of drugs i.e., solubility in water to that in DIF.
2. Materials and Methods 2.1. Experimental section 2.1.1. Chemicals
Soybean oil, kolliphor EL, acetonitrile (ACN), DMSO and Milli-Q water were obtained from Sigma-Aldrich (USA). Maisine 35-1 was from Gattefossé (France). Carvedilol was purchased from Molekula Group (Germany). Felodipine was obtained from Lundbeck A/S (Copenhagen).
2.1.2. Dog intestinal fluid samples
The DIF samples, used in the current project, were obtained from a previous study (Alskär et al., 2019). DIF collection was performed as follows: 2 g of LBF (not containing drugs) dispersed in 8 ml water, followed by 67 ml water, was administered to three Labrador dogs (Plura, Günther and Vanheden). Dog intestinal fluid (DIF) was sampled from duodenum stomas connected to plastic tubing for the duration of ≥1.5 h, and the time of collection and volume was noted. To inhibit further lipolysis of the formulation lipids, DIF was collected in vials standing on ice, and immediately treated with 1 μL of 1mM orlistat in ethanol. DIF sampling from stomas is erratic and depends on the time point of gastric emptying in each subject, i.e., exact time points for sampling could not be predetermined. Thus, the samples were divided into three approximate time periods: early around 0–5 min, mid around 5–20 min, and late around 20–90 min after administration for comparability of the results. The DIF
samples were stored in −80 °C until further analysis. DIF samples from only the mid time point were used in the current study.
2.1.3. Drug solubility in DIF after administration of 2 g LBF
DIF samples containing only 2 g LBF from the mid time point (5-20 min) were selected from all three dogs and were used. Solubility studies using a shake flask method were carried out and samples were analysed using HPLC-UV.
Aliquots of 300 µL DIF were added to Eppendorf tubes containing excess carvedilol and felodipine (4-5 mg), separately. Experiments were conducted in triplicates. The samples were vortexed and placed on a shaking plate (~200 rpm) in the incubator oven (Termaks Laboratory incubator) at 37 °C. After 24 h, the samples were centrifuged (37 °C, 2300g, 10 min), supernatant was removed and then vortexed, followed by 10-fold dilution in cold acetonitrile.
To remove precipitated proteins, the samples were spun a second time (4 °C, 2300g, 10 min), supernatant was removed and then vortexed. Prior to HPLC-UV analysis the samples were diluted again in mobile phase (10-fold). QC samples were also quantified in DIF to remove differences in matrices. QC samples in DIF were quantified only for one day, whereas QC samples in mobile phase were quantified on three different days. By conducting the experiment on several days, an estimate for the inter-day variability was established.
2.1.4. LBF type IIIA-LC preparation
The composition for the type IIIA-LC LBF used in this study, is shown in Table S1. 10 mg of LBF type IIIA-LC was prepared using 32.5% soybean oil, 32.5% maisine 35-1and 35%
kolliphor EL kept at 37℃. All proportions were measured in w/w.
2.1.5. Drug solubility in LBF dispersed in water
After 24h incubation of the prepared LBF (as above, 2.1.4), 1 g and 2 g LBF were dispersed in 75 mL ionic water, placed on a magnetic stirrer, and stirred for 20 min. About 5 mg of drug was weighed in 1.5 mL Eppendorf tubes. Aliquots of 300 µL of the dispersed LBF in water were added to Eppendorf tubes and incubated at 37 °C on a shaking plate for 24 h. An identical protocol for drug solubility in DIF containing 2 g LBF (as above, 2.1.3) was followed here, except for a six-fold dilution of the supernatant with mobile phase prior to HPLC-UV analysis.
2.1.6. HPLC-UV analysis
Analysis was performed with an HPLC (Agilent Technologies 1290 Infinity) with a Zorbax Eclipse XDB-C18 column (4.6×100 mm) (Agilent Technologies, US). The samples were quantified using HPLC-UV.
Stock solutions, of carvedilol and felodipine, for calibration curve (CC stock) and quality control (QC stock) were made in DMSO. Seven dilutions for carvedilol CC with a concentration range of 100-2.7 µg/mL, and three dilutions for QC with ACN: Milli-Q (1:1) were made (Appendix-Table S2, S3). The mobile phase consisted of sodium acetate buffer (NaAc; 25 mM, pH 5, 0.1 % triethylamine). A linear gradient with a flow rate of 1 ml/min over 4 min up to 20:80 (v/v) was used, followed by 2 min linear decrease back to 50:50. The injection volume was 20 μL, absorbance monitored at 243.4 nm, and the retention time of carvedilol was 2.0-2.2 min.
For felodipine, seven dilutions for CC ranging from 85-1.5 µg/mL, and three dilutions for QC were made with ACN: Milli-Q (1:1) (Appendix-Table S4, S5). MilliQ water was used as the mobile phase. The linear gradient with a flow rate of 1mL/min over 3 min up to 20:80 (v/v) was used, followed by 3 min linear decrease back to 50:50. The injection volume was 20µL, absorbance checked at 262.2 nm with a retention time of 3.9 min.
2.2. Computational section 2.2.1. Simulation systems of LBF
MD simulations were performed using GROMACS software version 2018.1(Lemkul and Lemkul, 2018). 3 local servers (48 CPUs each with GPU for fast water simulations, with Nvidia graphics cards) and quota on the supercomputer (Rackham, Uppmax, Uppsala, Sweden) were operated to run the simulations. Since GROMACS is parallelizable up to a great extent, we ran multiple simulations at a time (Hashemzadeh, Javadi and Darvishi, 2020).
Force field is an important component of classical molecular simulation (Brunsteiner, Khinast and Paudel, 2018). In this study, the Martini force field was used. The Martini coarse-grained (CG) model was originally developed for lipids (Marrink, De Vries and Mark, 2004). Martini uses a 4:1 mapping, i.e., on average four heavy atoms and associated hydrogens are represented by a single particle.
Simulations were carried out to examine the types of colloids formed with LBF in five different colloidal systems (Table 1), check drug-loading effects of both drugs and characterize the colloids formed. A colloid of 12 nm in diameter was simulated by using data from Alskär’s study (Alskär et al., 2019) (see Table S1). To diminish the effect of the initial configuration, all simulations were run in triplicate. Using the computational model of LBF, we formed five systems as shown in Table 1. These systems were then subjected to drug loading with 1, 10, 50 and 100 molecules, separately, to see if any interesting patterns can be observed. The idea behind analyzing the LBF in digested and undigested state was to get a better view on the effect of presence of BS and PL on the LBF, and to characterize the LBF in
digested and undigested state. Free energy change calculations were performed to evaluate how strongly solute molecules interacted with colloid, to indirectly compare it with experimentally observed solubility.
Table 1 Colloidal systems used for LBF characterization.
Colloidal systems Composition
LBF + water TG, DG, MG, Cremophor, water
Undigested LBF (1g sample) in DIF TG, DG, MG, Cremophor, water+ BS+ PL*
Undigested LBF (2g sample) in DIF TG, DG, MG, Cremophor, water+ BS+ PL*
Digested LBF (1g sample) in DIF MG, FFA, Cremophor, water+ BS+ PL*
Digested LBF (1g sample) in DIF MG, FFA, Cremophor, water+ BS+ PL*
*BS (sodium taurocholate), PL (Dipalmitoylphosphatidylcholine); For 2 g samples, average BS values of mid time point (22.7 ± 9.2 mM) were taken from Alskär et al’s study and PL values were 1/4th of BS. For 1g samples, values for BS and PL were half of the original values, respectively.
2.2.2. Simulation process
Using data from Alskär’s study (Table S1), LBF type IIIA LC was constructed using PACKMOL such that the diameter of the colloid was approximately 12 nm. This colloid was then put into a cubic box of length 45 nm and randomly solvated with water molecules. 2 g digested DIF contained 22.7 mM of BS and 5.675 mM of PL (1/4th of BS). 1 g digested DIF had half the values of BS and PL from 2 g digested DIF to mimic less digestion. For drug loading studies, the systems were loaded with 1, 10, 50 and 100 molecules of felodipine and carvedilol, separately and run parallelly (see Table 1). This was done to check if the drugs were distributed around the colloid or aggregated. Sodium (Na+) and chloride (Cl-) ions were added to the systems with a non-zero net charge. Energy minimization was done using steepest descent algorithm followed by NPT equilibration using Berendsen coupling and v- rescale with pressure set to 1 bar and temperature to 310 K (Pronk et al., 2013).
VMD was used to view the simulations and make graphical presentations. RDF was calculated to check where the components of LBF and drugs were distributed in the colloidal systems. It was calculated using the center of mass of each component of LBF, relative to the center of mass of the LBF. Radii of gyration were also calculated by selecting all the components of the LBF indexed as one group. Aspect ratios, defined as the ratio of the principal axis of revolution to the maximum diameter, were calculated to analyse the shape of the colloids formed. Average of triplicate runs was considered as the final radii of gyration for the system, but some runs were exempted due to lack of time.
Free energy calculations were also performed computationally, in order to investigate the energies of interactions between the solute molecule and the rest of the system. For correct calculations, only infinite dilution can be used, thus we always had only one API molecule in such simulations. A drug molecule was gradually removed from the simulation box by a stepwise reduction of the strength of its interactions with other molecules. The Hamiltonian derivative (change of the energy state of the system) was then calculated in the course of thermodynamic integration, via a method implemented in GROMACS software (gmx bar). A higher free energy change value (dG) means higher energy cost needed to take the solute away from the colloid, or, in other words, stronger interactions between the solute and the colloid. That can be indirectly related to solubility of the molecules. In case of a good correlation between dG and experimentally observed solubility, one could consider our computational model to be accurate and useful for further study of LBF-DIF-API interactions.
3. Results
3.1. Experimental section 3.1.1. Solubility studies of LBF
DIF samples (2.1.3) containing carvedilol showed almost thrice the solubility compared to felodipine (Figure 1A). Solubility of carvedilol was also three times higher than that of felodipine in both, 1g and 2g LBF dispersed in water (Figure 1B) The solubility was analyzed by measuring the concentration of the drug in mg/mL in DIF.
It was observed that both drugs showed higher solubility in DIF than LBF dispersed in water.
Despite the difference in matrices, solubility of carvedilol in 2 g LBF dispersed in water and DIF, was almost three times higher than that of felodipine (Appendix - Figure S5).
Figure 1 Solubility of carvedilol and felodipine in A) dog intestinal fluids containing 2g LBF and B) 1g and 2g LBF dispersed in 75 mL water. Values are depicted as mean ± SD, n = 3.
3.1.2. pH characterization of DIF
pH for DIF samples were measured with a Metrohm Biotrode. pH was measured (37 °C) at two time points; prior to addition of carvedilol and after ~20 h of incubation with the drug. pH did not have a significant difference on the solubility of carvedilol in DIF. The pH for DIF
A) B)
ranged between 6.4 and 7.6, mostly staying above 7.0. pH values did not vary much as shown in Table S6.
3.1.3. Drug solubility in DIF validation data
When DIF - QCs were compared to QCs in mobile phase, the results correlated with the CC in the range of 87% - 104%, which meant that there was acceptable loss of drug due to difference in matrices and the methods developed were accurate. QCs in the mobile phase showed relative standard deviation (RSD%) within ± 15%, as recommended by FDA, thereby showing precision (Fda and Cder, 2018). An inter-day variation, presented as RSD%, of 3% - 4% between QC samples of carvedilol was observed and 4% - 5% for felodipine. The calibration curves, for carvedilol and felodipine, showed R2 values of 0.99997 and 0.9984 (see Appendix, Fig S1, S2).
3.2. Computational section
3.2.1. LBF characterization and structure
Based on visualization of the colloids in VMD, as well as the RDF, it was found that there was a certain order to the distribution of components (Figure 4, Figure 5, Table 2). For the pure LBF in water, from the center outwards, the trend is as such: triglycerides (TG) > diglycerides (DG) >
monoglycerides (MG) > cremophor (CRE).
The undigested LBF was then simulated in DIF (Table 1), by adding bile salt (TAUR) and phospholipid (DIPC). The TAUR and DIPC tend to be located at the surface of the colloid (Figure 4). To simulate the digested LBF every TG and DG was replaced with MG and free fatty acids (FFA) in the presence of TAUR and DIPC, modelling the intestinal fluid of a dog (Figure 4, 5).
The MG and FFA seem to distribute a little away from the center. CRE makes an intermediate shell, with some contacts with water remaining. MG and FFA are then both at the surface and in the core.
RDF allowed us to observe the distribution of components within components of colloid - TG, DG and MG (Figure 5). The order of RDF remained the same regardless of the type of glyceride:
Stearic acid > Oleic acid ~ Palmitic acid ~ Linoleic acid > Linolenic acid. CRE was always found to be distributed about 5.5 nm from the center of the colloid in all colloidal systems.
Based on analysis, while the LBF underwent digestion, the radii of gyration increased from 4.87 to 5.18 nm as shown in Table 2. Both undigested LBF (1 g and 2 g), had the same amount of LBF but only the DIPC and TAUR values were changed – decreased to mimic 1 g and increased to mimic 2g, and estimate the effect of the ratio of LBF, and DIF. Aspect ratios were a little over 1, stating that the shape of the colloid remained almost spherical.
Figure 4 Graphical representation of A) Undigested LBF in water, B) 2g LBF in digested DIF, C) 1g LBF in undigested DIF and, D) 1g LBF in undigested DIF.
Figure 5 RDF of individual components of undigested pure LBF in water. A) Monoglycerides, B) Diglycerides, C) Triglycerides. Cremophor is left in all panels as a reference for comparison.
3.2.2. Drug loading
All five colloidal systems (see Table 2) were loaded with felodipine in the order of 1, 10, 50 and 100 molecules. Carvedilol drug loading results are displayed in the appendix (Table S8). As can be seen in Table 3, colloids could accommodate every felodipine molecule at all studied drug loadings. It was also observed that the shape of the digested LBF changes as the drug loading increases owing to change in aspect ratios (Appendix - Table S7). Carvedilol loading in colloids resulted in the drug self-assembling itself (Appendix- Table S8).
B
C A
Table 2 Characterization of LBF with radii of gyration and aspect ratios.
Colloidal system
Pictorial representation (VMD)
Diameter (nm)
Radius of gyration
(nm)
Aspect ratio
2 g LBF + H2O
12.630 4.87 1.03
1g LBF
undigested* 12.603 4.92 1.03
2g LBF
undigested 12.620 5.00 1.03
1g LBF
digested* 12.598 5.12 1.05
2g LBF
digested 12.608 5.18 1.05
*CRE = kolliphor (cremophor), MG = monoglycerides, TG = triglycerides, DG = diglycerides, FEL = felodipine, FFA = free fatty acids, DIPC = dipalmitylphosphotidylcholine, TAUR = sodium taurocholate,
-MG -TG -DG -FFA -DIPC -TAUR
-CRE
Table 3 Drug loading of Felodipine in colloidal systems with 1g and 2g undigested LBF.
Drug loading
(molecules) LBF + H2O
2g undigested LBF + DIF
2g digested LBF + DIF
1
10
50
100
*Red molecules are felodipine drug molecules.
Radii of gyration within triplicates did not seem to vary significantly (Appendix - Table S7).
Free energy calculations were also carried out and did not vary between five colloidal structures as shown in Figure 6. However, carvedilol interacted more strongly with colloids than felodipine. dG values for carvedilol in 2g LBF digested in the presence of DIF were slightly higher than 1g and, than in water, which is in line with experimental observations.
RDF analysis was also done to check where the components of LBF lie within the colloid, and
where the drugs tend to re-distribute in different colloids. As shown in Figure 7, digestion had a no visible effect on drug distribution. It was also observed that, after digestion the structure of the colloid changed (see Figure 4) and showed two distinct layers of MG and FFA, separated by an intermediate shell of CRE. The molecules of BS and PL slightly increased the size of the colloid and were located at the surface of the colloid.
Figure 6 Free energy calculations for colloidal systems. SD is within 2% of the presented average values.
Figure 7 Radial distribution function of components of LBF for all five colloids, as marked above the panels.
Digestion had a strong effect on the distribution of the colloidal components, splitting MG and FFA into two layers. It also affected the positions of the felodipine molecules. Carvedilol did not show significant changes in the radial distribution between the systems.
4. Discussion
We have compared the performance of LBF as a PWSD solubilizer in water (typical in vitro study), and in dog intestinal fluid. As seen from the experimental study, DIF improves the solubility of both drugs (Figure 1A, B). Apart from that, increased amounts of LBF also showed better solubility, which leads to a hypothesis that the solubilization of the drugs is limited by the amount or size of the colloids. This can be linked to the presence of more digestion of LBF, thereby increasing hydrophobic capacity (Feeney et al., 2016; Alskär et al., 2018). It is not yet clear whether surfactants and co-solvents are necessary parts for some of the drugs, or it is more food-relevant components, such as glycerides and free fatty acids, that can also be stable and lead to the same solubilization effect at certain proportions. The solubility of carvedilol was found to be three times higher than that of felodipine in all dogs.
The effect of adding FFA has been shown to give a more pronounced solubility effect, compared to increased BS levels, for cinnarizine in simulated intestinal fluid (Larsen et al., 2013). Permeation studies could be carried out to check the permeation of these drugs since it has been previously studied that a lipophilic drug typically partitions into the colloidal structures formed in the presence of solubilizing compounds (e.g., lipids and free fatty acids), reducing the total drug available for permeation (Keemink, Mårtensson and Bergström, 2019).
From computational simulations, we observed that all 12 nm colloids (in water and DIF, digested and undigested) could easily accommodate 100 molecules of felodipine though some evident aggregation was observed in the case of carvedilol (Appendix Table S8). It was also observed that due to a certain repeating trend in the distribution of fatty acids, the length of the tails and the positions of double bonds, could have a potential effect on solubilization (Figure 5). Both, digested and undigested 1 g colloids showed a larger diameter compared to 2 g colloids (Table 2). TAUR and DIPC were seen to distribute on the periphery of the colloid which could be due to weak base like property of TAUR and amphiphilic property of DIPC and can be potentially used for the drugs that could be screened from the water later within the inner shell. This shows that highly hydrophobic molecules will most likely not interact with them.
Based on analysis, when the LBF underwent digestion, the radii of gyration increased from 4.87 to 5.18 nm as shown in Table 2.The spheroid shape might be due to higher degrees of freedom of the new system, thereby leading to an increased entropy and imperfection in packing. Thus, we propose that LBF composed of smaller glycerides and free fatty acids can serve better for the purpose of solubilization of the studied drugs. For the lipophilic drugs, components of intestinal fluid should play a bigger role, but for 1 API molecule no significant
difference in dG was observed between five colloids (Figure 6).It might be related to the fact that only one drug molecule was introduced for the simulations, which is dictated by theoretical considerations. For just one molecule all 12 nm colloids served equally as an energetically favorable environment. Nevertheless, dG for systems in the presence of DIF was generally higher than LBF in water. Apart from that, the energy difference was higher for carvedilol than for felodipine, which is in qualitative agreement with experimental data, as carvedilol had a higher solubility and was better solubilized. Digestion of LBF did not significantly affect the distribution of felodipine molecules. Carvedilol did not show any significant changes in either of the colloidal systems and was seen to self-assemble (Table S8).
There is an enormous scope to improve future research, but we believe that the qualitative match between experimental and computational data confirms that our in-silico model is valid.
5. Conclusion
In this work we studied solubility of carvedilol and felodipine in LBF in water and LBF in DIF and compared it with computational data. As was observed experimentally, the presence of DIF improved the solubility of drugs significantly. Digestion and amount of LBF were found to be important factors for the drugs’ solubilization. The computational data showed a good potential for in-depth studies on the mechanisms of solubilization, improvement and development of new LBFs. However free energy calculation only reflected some of the trends, as a single drug molecule was always very well solubilized within big micelles. Therefore, a further improvement of computational approach would be beneficial for the study. It would also be of great interest to investigate systems at critical drug loadings, at which APIs would start to self-aggregate either within, on the colloids or outside. An additional study on colloids of varying sizes and effect of fatty acid chain lengths could complement the current research and provide more insight on the effect of LBF on solubilization.
6. References
Alskär, L. C. et al. (2018) ‘Impact of Drug Physicochemical Properties on Lipolysis- Triggered Drug Supersaturation and Precipitation from Lipid-Based Formulations’. doi:
10.1021/acs.molpharmaceut.8b00699.
Alskär, L. C. et al. (2019) ‘Effect of lipids on absorption of carvedilol in dogs: Is
coadministration of lipids as efficient as a lipid-based formulation?’, Journal of Controlled Release, 304, pp. 90–100. doi: 10.1016/j.jconrel.2019.04.038.
Alves, J. M. V., Prado, L. D. and Rocha, H. V. A. (2016) ‘Evaluation and correlation of the physicochemical properties of carvedilol’, Pharmaceutical Development and Technology, 21(7), pp. 856–866. doi: 10.3109/10837450.2015.1073740.
Brunsteiner, M., Khinast, J. and Paudel, A. (2018) ‘Relative contributions of solubility and mobility to the stability of amorphous solid dispersions of poorly soluble drugs: A molecular dynamics simulation study’, Pharmaceutics, 10(3). doi: 10.3390/pharmaceutics10030101.
Buckley, S. T. et al. (2013) ‘Biopharmaceutical classification of poorly soluble drugs with respect to “enabling formulations”’, European Journal of Pharmaceutical Sciences. Elsevier B.V., pp. 8–16. doi: 10.1016/j.ejps.2013.04.002.
Fda and Cder (2018) Bioanalytical Method Validation Guidance for Industry Available at:
http://www.fda.gov/Drugs/GuidanceComplianceRegulatoryInformation/Guidances/default.ht mand/orhttp://www.fda.gov/AnimalVeterinary/GuidanceComplianceEnforcement/Guidancefo rIndustry/default.htm (Accessed: 25 May 2021).
Feeney, O. M. et al. (2016) ‘50 years of oral lipid-based formulations : Provenance, progress and future perspectives’, Advanced Drug Delivery Reviews, 101, pp. 167–194. doi:
10.1016/j.addr.2016.04.007.
Hashemzadeh, H., Javadi, H. and Darvishi, M. H. (2020) ‘Study of Structural stability and formation mechanisms in DSPC and DPSM liposomes: A coarse-grained molecular dynamics simulation’, Scientific Reports, 10(1), pp. 1–10. doi: 10.1038/s41598-020-58730-z.
Kalepu, S. and Nekkanti, V. (2015) ‘Insoluble drug delivery strategies: Review of recent advances and business prospects’, Acta Pharmaceutica Sinica B. Chinese Academy of Medical Sciences, pp. 442–453. doi: 10.1016/j.apsb.2015.07.003.
Keemink, J., Mårtensson, E. and Bergström, C. A. S. (2019) ‘Lipolysis-Permeation Setup for Simultaneous Study of Digestion and Absorption in Vitro’, Molecular Pharmaceutics, 16(3), pp. 921–930. doi: 10.1021/acs.molpharmaceut.8b00811.
Larsen, A. T. et al. (2013) ‘Bioavailability of cinnarizine in dogs: Effect of SNEDDS loading level and correlation with cinnarizine solubilization during in vitro lipolysis’, Pharmaceutical Research, 30(12), pp. 3101–3113. doi: 10.1007/s11095-013-1145-x.
Larsson, P., Alskär, L. C. and Bergströ, C. A. S. (2017) ‘Molecular Structuring and Phase Transition of Lipid-Based Formulations upon Water Dispersion: A Coarse-Grained Molecular Dynamics Simulation Approach’. doi: 10.1021/acs.molpharmaceut.7b00397.
Lemkul, J. and Lemkul, J. A. (2018) ‘From Proteins to Perturbed Hamiltonians: A Suite of Tutorials for the GROMACS-2018 Molecular Simulation Package [Article v1.0]. doi:
10.33011/livecoms.1.1.5068.
Marrink, S. J., De Vries, A. H. and Mark, A. E. (2004) ‘Coarse Grained Model for
Semiquantitative Lipid Simulations’, Journal of Physical Chemistry B, 108(2), pp. 750–760.
doi: 10.1021/jp036508g.
Pouton, C. W. and Porter, C. J. H. (2008) ‘Formulation of lipid-based delivery systems for oral administration: Materials, methods and strategies’, Advanced Drug Delivery Reviews.
Adv Drug Deliv Rev, pp. 625–637. doi: 10.1016/j.addr.2007.10.010.
Pronk, S. et al. (2013) ‘GROMACS 4.5: A high-throughput and highly parallel open source molecular simulation toolkit’, Bioinformatics, 29(7), pp. 845–854. doi:
10.1093/bioinformatics/btt055.
Rodriguez-Aller, M. et al. (2015) ‘Strategies for formulating and delivering poorly water- soluble drugs’, Journal of Drug Delivery Science and Technology, 30, pp. 342–351. doi:
10.1016/j.jddst.2015.05.009.
Xiao, L. et al. (2016) ‘The in Vitro Lipolysis of Lipid-Based Drug Delivery Systems: A Newly Identified Relationship between Drug Release and Liquid Crystalline Phase’, BioMed Research International, 2016. doi: 10.1155/2016/2364317.
7. Appendix
Table S1 Composition of Type IIIA-LC lipid formulation.
Name Excipient type
% composition
Description
Soybean oil LCTG 32.5
FAs: 10% palmitic acid (C16:0), 7% linolenic acid (C18:3), 53% linoleic acid (C18:2), 25%
oleic acid (18:1), 5% stearic acid (18:0) Maisine 35:1 LCMIX 32.5
Blend of 45% MG, 46% DG, 9% TG. FAs:
11% palmitic acid (C16:0), 1% linolenic acid (C18:3), 55% linoleic acid (C18:2), 31% oleic
acid (18:1), 2% stearic acid (18:0) Kolliphor
EL/ELP Surfactant 35 Polyethoxylated castor oil, HLB 12–14
*Long-chain triglyceride (LCTG), long-chain mixed mono-, di-, triglyceride (LCMIX)
Table S2 Carvedilol calibration curve dilutions Dilution
factor
CC stock (µL) Mobile phase (µL)
Final volume (µL)
Final
concentration (µg/mL)
50 980 20 1000 100
100 990 10 1000 50
200 900 300 from 50X 1000 30
400 900 300 from 100X 1000 15
800 900 300 from 200X 1000 9
1600 900 300 from 400X 1000 4.5
3200 900 300 from 800X 1000 2.7
Table S3 Carvedilol quality control dilutions
Quality controls Stock 10X (QC) Mobile phase (µL)
Final volume (µL)
Final concentration
(µg/mL)
QC 1 20 980 1000 10
QC 2 80 920 1000 40
QC 3 160 840 1000 80
Table S4 Felodipine calibration curve dilutions
Dilution Mobile Phase
(µL)
CC 10X (µL) Concentration (µg/mL)
625 998 2 1,5
333,33 997 3 2,25
100 990 10 5
50 980 20* 10
25 960 40* 30
9.0909 890 110* 55
5.882 830 170* 85
*The respective dilutions were made from another CC 10X (500µg/mL) to adhere to the linearity for the calibration curve.
Table S5 Felodipine quality control dilutions Quality controls QC 10x (µL)
(500µg/mL)
QC stock
(5000µg/mL)
Mobile phase (µL)
Conc (µg/mL)
QC1 12 -- 988 6
QC2 -- 4 996 20
QC3 -- 16 884 80
Table S6 pH characterization of DIF before and after addition of carvedilol.
Dogs Before After
Plura 7,6 7
Vanheden 6,8 7,1
Günther 6,4 6,9
Table S7 Radii of gyration and aspect ratios for all systems with Felodipine
Systems
Rg (radius of
gyration)
Aspect
ratios SD CV% No. of
molecules
LBF in water
4.87 1.03 0.13 3% 100
4.90 1.03 0.00 0% 50
4.82 1.03 0.09 2% 10
4.77 1.03 0.09 2% 1
1g Undigested
5.02 1.03 0.02 0% 100
4.96 1.03 0.01 0% 50
4.94 1.03 0.00 0% 10
4.93 1.03 0.01 0% 1
2g Undigested
5.09 1.03 0.01 0% 100
5.06 1.03 0.06 1% 50
5.00 1.03 0.01 0% 10
5.00 1.03 0.03 1% 1
1g Digested
5.27 1.06 0.12 2% 100
5.18 1.05 0.05 1% 50
5.12 1.05 0.00 0% 10
5.12 1.06 0.00 0% 1
2g Digested
5.28 1.05 0.06 1% 100
5.21 1.20 0.02 0% 50
5.15 1.05 -- -- 10
5.16 1.11 0.00 0% 1
Figure S1 Calibration curve for felodipine
Figure S2 Calibration curve for carvedilol.
Figure S3 Chromatogram of Carvedilol in DIF
Figure S4 Chromatogram of felodipine in DIF
Figure S5 Average solubilities of felodipine and carvedilol in a) dog intestinal fluid containing 2g LBF ( ) and b) 2g LBF dispersed in 75 mL water ( ); average solubility of both drugs is presented here along with standard deviation as error brackets.
Table S8 Carvedilol drug loading in 1g undigested and digested LBF.
Drug loading (molecules)
1 g undigested LBF
Rg Aspect
ratio 1 g digested
LBF
Rg Aspect ratio
1 4.94 1.03 1.06 5.14
10 5.00 1.03 1.07 5.46
50 5.16 1.04 Results were inconclusive. -- ---
100 5.91 1.06 Results were inconclusive. -- ----
