bs_bs_banner

Journal of Pharmacy
Research Paper
And Pharmacology

A simplified method to screen for in-vivo performance of

oral lipid formulations
Murat Kilic and Jennifer Dressman
Institute of Pharmaceutical Technology, Goethe University, Frankfurt am Main, Germany

Keywords Abstract
biorelevant media; drug precipitation; in-vitro
lipolysis; lipid formulations; poorly soluble Objectives To develop a simplified in-vitro screening method for oral lipid-based
drugs formulations using intestinal biorelevant media including lipolysis, as an alterna-
tive to pH-stat lipolysis models.
Correspondence
Method Fasting state simulated intestinal fluid version 2 (FaSSIF-V2) and early
Jennifer Dressman, Institute of Pharmaceutical
Technology, Goethe University, Max-von-Laue
fed state simulated intestinal fluid (FeSSIF) were modified for use in a simplified
Straße 7, 60438, Frankfurt am Main, lipolysis screening method. This screening method consists of the following steps:
Germany. dispersion of the lipid formulation in biorelevant media; incubation of disper-
E-mail: dressman@em.uni-frankfurt.de sions on an orbital shaker for 60 min at 37°C; ultracentrifugation of dispersions
and drug assay of supernatants. This method was evaluated using four lipid-based
Received August 19, 2013 formulations containing danazol, which had previously been assessed by in-vitro
Accepted October 15, 2013
pH-stat lipolysis and compared in an in-vivo study in dogs.
doi: 10.1111/jphp.12182
Key findings Biorelevant media were modified under consideration of both
physiological and practical aspects, including adjustment of the pH to 6.5, the
addition of calcium ions and the addition of 100 U/ml porcine pancreatin to
enable lipolysis of a test formulation. Using a modified FaSSIF-V2, the same rank
order in performance of four danazol formulations as previously observed in a
pH-stat model was observed, and these results also reflected the in-vivo study
results. The results in modified early FeSSIF suggested that there would be a
change in the rank order of formulation performance in the fed state compared
with the fasted state. By comparing the formulation behaviour in the presence and
absence of pancreatin, it was concluded that dispersion is more important than
lipolysis for precipitation from the formulation in the fasted state, but that
lipolysis is predicted to increase in relevance in the fed state.
Conclusion The new, simplified method for lipolysis enables a more efficient
screening for the in-vivo performance of lipid formulations in the fasted state and
enables a prediction of formulation behaviour in both the fed and fasted states. An
additional advantage of the method is that the relative influence of lipolysis and
dispersion on drug release can be directly compared.

Introduction
Oral lipid formulations are one of several promising
approaches to improve the bioavailability of poorly soluble
drugs.[1] As long as the active pharmaceutical ingredient
(API) is dissolved in the lipid vehicle, the dissolution step
for the API from the dosage form is circumvented, and
absorption can occur as soon as the API disperses/partitions
into the gastrointestinal fluids or is released by digestion of
the vehicle. Most lipid formulations that have been recently
developed belong to the group of self-emulsifying formula-
tions.[2] After oral intake, these formulations form an emul-
sion or a micro-emulsion upon dilution with the gastric
juices, leading to improved drug release and absorption.[3–5]
In addition to dispersion of the dosage form in the gas-
trointestinal fluids, a second important consideration in the
in-vivo performance of lipid formulations is the possibility
of gastrointestinal lipolysis.[6,7] Although digestion of the
lipid vehicle is deemed crucial to drug release from simple
lipid formulations (i.e. Type I formulations according the

© 2013 Royal Pharmaceutical Society, Journal of Pharmacy and Pharmacology, 66, pp. 615–623 615
Screening oral lipid formulations Murat Kilic and Jennifer Dressman

Lipid Formulation Classification System (LFCS)[6,7]), Priest, France. Sodium taurocholate (NaTC, 97% pure) was
lipolysis could in fact have a negative impact on API obtained from Prodotti Chimici e Alimentari SpA,
bioavailability in the case of self-emulsifying formulations Basaluzzo, Italy. Glyceryl monooleate (Rylo M19 Pharma®)
since degradation of the emulsifiers could lead to a decrease was a gift from Danisco Specialities, Brebrand, Denmark.
in their ability to solubilise the API and thus to precipita- Lecithin (Lipoid E PC®, 99.1% pure) was donated from
tion of the API.[2,8,9] These contrasting but important effects Lipoid GmbH, Ludwigshafen, Germany. Sodium oleate
of lipolysis on lipid formulation behaviour led to the devel- (82.7% pure) was obtained from Riedel-de Haën, Seelze,
opment of in-vitro lipolysis models as the main tool for the Germany. Danazol (min 98%), maleic acid (99% pure) and
prediction of in-vivo performance of lipid formulations. pancreatin (8 × United States Pharmacopeia (USP) specifi-
Typically, models to address lipolysis and its impact on drug cations) were purchased from Sigma-Aldrich Chemie
release/solubilisation are based on a pH-stat set-up. The GmbH (Steinheim, Germany). Acetonitrile, calcium chlo-
set-up generally consists of a temperature-controlled vessel ride (CaCl2), dichloromethane, sodium chloride (NaCl),
(at 37°C) and an aqueous buffer medium adjusted to a NaOH pellets, hydrochoric acid and ethanol were all of ana-
near-neutral pH (typically 6.5 or 7.5), which contains bile lytical grade and purchased from Merck KGaA (Darmstadt,
acids (e.g. sodium taurocholate) and phospholipids (e.g. Germany).
lecithin), calcium chloride, and porcine pancreatin. The
calcium is added to sequester the free fatty acids produced Biorelevant media for lipolysis screening
during lipolysis and can be added in different ways: one
Based on physiological and practical considerations,
being to add about 5 mm calcium ion to the medium before
FaSSIF-V2 and early FeSSIF were modified slightly to simu-
beginning the lipolysis, and the other being to add the
late preprandial and postprandial states of the upper duo-
calcium ion continuously during the test (usually at a rate
denum in the test set-up. Tables 1 and 2 compare the
between 0.045 and 0.181 mmol/min; this approach has
original media with the modified media applied in this
been termed the ‘dynamic lipolysis model’).[10–12]
The pH-stat models described are associated with
implicit disadvantages because of their experimental design; Table 1 Compositions of FaSSIF-V2[13] and the modified FaSSIF-V2
the conditions are only partly representative of the intesti- used for lipolysis screening
nal physiology, and the methods are low throughput. There- Modified FaSSIF-V2
fore, it would be desirable to develop a new, simplified Composition FaSSIF-V2 for lipolysis screening
screening method for lipolysis effects. Such a method
Sodium taurocholate (mM) 3 3
should use a more physiologically relevant set of test condi- Lecithin (mM) 0.2 0.2
tions, be able to screen for lipolysis effects in the fed as well Maleic acid (mM) 19.12 28.6
as the fasted state, and be simpler to use than the pH-stat Sodium hydroxide (mM) 34.8 52.5
methods and therefore enable a higher thoughput. Sodium chloride (mM) 68.62 68.62
This study was therefore aimed at developing a simplified Pancreatin (U/ml) – 100
Calcium chloride (mM) – 5
lipolysis screening method based on two biorelevant media:
pH 6.5 6.5
fasting state simulated intestinal fluid version 2 (FaSSIF-V2)
Buffer capacity (mmol/ΔpH/L) 10 25
simulating the fasted state and ‘early fed state simulated
intestinal fluid (FeSSIF)’ simulating the fed state.[13] Model
lipid formulations containing the poorly soluble API Table 2 Compositions of early FeSSIF[13] and the modified early FeSSIF
danazol, for which both in-vitro and in-vivo data had been used for lipolysis screening
previously published in the literature,[8] were used to dem-
Modified early FeSSIF for
onstrate that the proposed lipolysis screening method is Composition early FeSSIF lipolysis screening
useful for discriminating between dispersion and lipolysis
effects on the one hand and for predicting the rank order of Sodium taurocholate (mM) 10 10
Lecithin (mM) 3 3
lipid formulation performance on the other hand.
Glyceryl monooleate (mM) 6.5 6.5
Sodium oleate (mM) 40 0.8
Materials and Methods Maleic Acid (mM) 28.6 28.6
Sodium hydroxide (mM) 52.5 52.5
Chemicals
Sodium chloride (mM) 145.2 145.2
Soybean oil (refined, Ph.Eur.) and Cremophor EL Pancreatin (U/ml) – 100
(macrogolglycerol ricinoleate, Ph.Eur.) were purchased Calcium chloride (mM) – 5
pH 6.5 6.5
from Caelo, Hilden, Germany. Maisine® 35-1 (primarily
Buffer capacity (mmol/ΔpH/L) 25 25
glycerol monolinoleate) was donated from Gattefosse, St.

616 © 2013 Royal Pharmaceutical Society, Journal of Pharmacy and Pharmacology, 66, pp. 615–623
Murat Kilic and Jennifer Dressman
study and the modifications and their rationale are dis-
cussed in detail in the Results section.
Preparation of media
Media were prepared based on the methods described by
Jantratid and colleagues.[13] Briefly, to prepare 1000 ml of
modified FaSSIF-V2 without pancreatin, a buffer at pH 6.5
consisting of 4.01 g sodium chloride, 3.32 g maleic acid, 2.1 g
sodium hydroxide and approximately 900 ml demineralised
water in a volumetric flask was initially prepared. Approxi-
mately 100 ml of this buffer was transferred into a round
bottom flask, and 1.67 g sodium taurocholate was dissolved
in the buffer. To prepare the concentrated lecithin solution,
3.15 g lecithin was dissolved in 10 ml dichloromethane.
0.5 ml of this concentrate was dispersed in the buffer solu-
tion containing the bile salt, which resulted in the formation
of an emulsion. The dichloromethane was then removed
stepwise using a rotary evaporator at a temperature of 40°C
and applying a vacuum. In the final step, a 100 mbar vacuum
was applied for at least 15 min to result in a micellar solution
with no perceptible odour of dichloromethane. After allow-
ing the micellar solution to cool down to ambient tempera-
ture, it was transferred back into the remaining bile salt
solution. After dissolving 0.56 g calcium chloride into the
solution, the pH was adjusted to pH 6.5 with either diluted
hydrochloric acid or sodium hydroxide solution, and the
volume was adjusted to 1 l.
To prepare modified FaSSIF-V2 with pancreatin, the
same procedure was followed, with the exception that 20 ml
of an aqueous pancreatin extract (described later) was
added just before the pH and volume adjustment steps.
To prepare 1000 ml of modified early FeSSIF without pan-
creatin, a buffer at pH 6.5 consisting of 8.49 g sodium
chloride, 3.32 g maleic acid, 2.1 g sodium hydroxide and
approximately 900 ml demineralised water in a volumetric
flask was initially prepared. Approximately 100 ml of this
buffer was transferred into a round bottom flask, and 1.67 g
sodium taurocholate was dissolved into the buffer. To prepare
the concentrated lecithin solution, 3.15 g lecithin was first
dissolved in 10 ml dichloromethane. After dispersing 8 ml of
this concentrate in the bile salt solution an emulsion was
obtained. The dichlormethane was then removed stepwise as
described for FaSSIF-V2 earlier. In this case, also, the pro-
cedure resulted in a micellar solution with no perceptible
odour of dichloromethane. After the micellar solution had
returned to ambient temperature, 7 ml of a solution of 3.66 g
glycerol oleate in 10 ml dichloromethane was added, again
resulting in an emulsion. The dichloromethane was removed
stepwise as described earlier and also resulted in a micellar
solution with no perceptible odour of dichloromethane. After
cooling this micellar solution back down to ambient tem-
perature, 0.24 g sodium oleate was dissolved into the solu-
© 2013 Royal Pharmaceutical Society, Journal of Pharmacy and Pharmacology, 66, pp. 615–623
Screening oral lipid formulations
tion, and the solution was transferred back to the remaining
pH 6.5 buffer. After dissolving 0.56 g calcium chloride in the
buffer solution containing the lipid components, the pH was
re-adjusted to pH 6.5 if necessary using diluted hydrochloric
acid or sodium hydroxide solution, and the volume was
adjusted to 1 l.
To prepare modified early FeSSIF with pancreatin, the
same procedure was followed, with the exception that 20 ml
of an aqueous pancreatin extract (described next) was
added before the pH and volume adjustment steps.
To prepare 20 ml pancreatin extract, 6.25 g porcine
pancreatin (8 × USP specification) was mixed with
demineralised water until homogenously dispersed. The
resulting suspension was centrifuged at 5°C and 4000 rpm
for 15 min (5804 R Eppendorf laboratory centrifuge with
rotor F45-30-1 from Eppendorf AG, Hamburg, Germany),
after which the supernatant was removed and used for
preparation of the screening media.
Lipolysis screening method
Experimental set-up
Figure 1 shows the experimental set-up of the lipolysis
screening method. To run the lipolysis screen, 10 ml of the
appropriate medium is added to a 20-ml scintillation vial,
and 100 mg of the formulation is added to the medium.
The lid is then screwed on tight, and the system is vortexed
until the formulation is visually well-dispersed in the
medium. The dispersion is then placed on an orbital shaker
(Heidolph Polymax 1040) at 37°C and incubated for
60 min. After the incubation, the dispersion is vortexed to
ensure redispersion, then an 8-ml sample is removed and
placed in a centrifuge tube (thick wall, polycarbonate,
10.5 ml, 13 × 89 mm from Beckman Coulter GmbH,
Krefeld, Germany). The sample is then centrifuged for
60 min at 37°C and 40 000 rpm in an ultracentrifuge (ultra-
centrifuge Optima™ L 80 with rotor Type 70.1 Ti from
Beckman Coulter GmbH) to separate the fluid phases from
the solid phase, which contains any API and calcium soaps
that have precipitated. The fluid phase is then removed
quantitatively using a pipette. After ensuring that the fluid
phase is well mixed, a 300-μl sample is transferred to a cen-
trifugation tube and diluted with 600 μl mobile phase
(acetonitrile 80% and water 20%). Subsequently, this dis-
persion is centrifuged for 15 min at 14 000 rpm and
ambient temperature (5804 R Eppendorf laboratory centri-
fuge with rotor F45-30-1 from Eppendorf AG), and the
supernatant is analysed for the API by HPLC.
Assessment of suitability of the screening method
To assess the suitability of the lipolysis screening method, it
was applied to four lipid formulations containing the
617
Screening oral lipid formulations Murat Kilic and Jennifer Dressman

drug assay

1 h, 37°C

phases after
ultracentrifugation
ultracentrifugation

precipitate
dispersion
and incubation

Figure 1 Illustration of the experimental set-up for the lipolysis screening method.

Danazol plasma concentration (ng/ml) 140

% of the initial drug concentration

120 F1
120
100 100
in the aqueous phase

F1 F2
80 80
F3
60 60
F2
F4
40 F3 40

20 F4 20

0 0
0 10 20 30 40 50 60 0 2 4 6 8 10
Time (min) Time (h)

Figure 2 Comparison of in-vitro lipolysis results with plasma profiles of danazol in beagle dogs after administering four lipid formulations in the
fasted state (modified from figures originally published by Cuine et al.[8] and modified in Porter et al.,[3] reproduced with permisson).

poorly soluble API danazol that had been previously tested Preparation of four lipid-based formulations
by Cuine and colleagues.[8] In that study the formulations of danazol
were first compared according to a plot of % drug remain-
The compositions of the four danazol formulations are
ing in solution as a function of time obtained in a
given in Table 3. They were prepared as described in the
standard in-vitro pH-stat model. In addition, an in-vivo
original publication.[8]
pharmacokinetic study of the four formulations was con-
ducted in fasted beagle dogs. The results are reproduced in
Solubility studies
Figure 2. In their paper, Cuine et al. compared the in-vitro
with the in-vivo results and concluded that lipolysis plays a To determine the apparent solubility of danazol in
major role in the precipitation of danazol in Beagle dogs.[8] biorelevant media, a miniaturised shake flask method was
A comparison of the results from the new screening employed.[14] An excess amount of danazol was weighed
method with the Cuine data should therefore enable a first into a Whatman Uni-Prep® filter chamber (0.45 μm
assessment of the new lipolysis screening model. Experi- polytetrafluoroethylene). Then, 3 ml of the test media was
ments were conducted in triplicate. added, and the chamber was closed with the corresponding

618 © 2013 Royal Pharmaceutical Society, Journal of Pharmacy and Pharmacology, 66, pp. 615–623
Murat Kilic and Jennifer Dressman
Table 3 Composition of the four danazol formulations from Cuine et al.[8]
Formulation Danazol in mg/mla
F1 11.8
F2 16.1
F3 17.8
F4 18.0
a
These concentrations represent 80% of saturated solubility of danazol in the respective vehicle.
plunger. The samples were shaken at 37°C in an orbital
shaker (Heidolph Polymax 1040) for 24 h. Finally, samples
were filtered by pressing the plunger into the medium and
analysed by HPLC after appropriate dilution with mobile
phase. Experiments were conducted in triplicate.
The HPLC system
For the quantitative analysis of danazol in samples obtained
from solubility and lipolysis studies an isocratic reversed-
phase HPLC system (LaChrom® L-7110 pump, a LaChrom
L-7400 UV–Vis-Detector, a LaChrom L-7200 autosampler
from Merck Hitachi and EZ-Chrom Elite Data System Soft-
ware® V 3.3.1 from Biochrom Ltd, Cambridge, UK). A
LiChroCART® RP-18 5 μm, 125 × 4 mm analytical column
(Merck) was applied. The mobile phase consisted of 80%
acetonitrile and 20% ultra-purified water (Milli-Q®). The
flow rate was set at 1.0 ml/min resulting in elution of
danazol at approximately 3.5 min. The detection wave-
length was at 280 nm. This method is based on the method
published by Schamp.[15]
Statistical methods
All experiments were performed in triplicate. Standard
deviations were calculated using Microsoft Excel 2007
(Microsoft Corporation, Redmond, Washington, USA), and
the error bars in figures represent standard deviations in all
cases. The software SigmaPlot© for Windows Version 11.0
from Systat Software, Inc. (San Jose, California, USA) was
used for statistical processing. For the screening method for
the effect of formulations on danazol, solubilisation was
evaluated using the Kruskal–Wallis one-way analysis of
variance test. In all cases, post-hoc comparisons of the
means of individual groups were performed using pairwise
multiple comparison procedures (Bonferroni t-test). A
result of P < 0.05 denoted a significant difference in all
cases.
Results and Discussion
Design of media for lipolysis screening
As part of their update of biorelevant media, Jantratid et al.
had recommended FeSSIF-V2 with pancreatin and calcium
© 2013 Royal Pharmaceutical Society, Journal of Pharmacy and Pharmacology, 66, pp. 615–623
Screening oral lipid formulations
Lipid component (soybean
oil : maisine 35-1; 1 : 1) (%) Cremophor EL (%) Ethanol (%)
60 30 10
37.5 55 7.5
18 64 18
0 65 35
chloride for release testing of lipid formulations.[13] Based
on their work, biorelevant media were modified for the
simulation of lipolysis in both the preprandial and post-
prandial intestine in this study. FaSSIF-V2 und early FeSSIF
represent a suitable basis for meeting both the practical and
physiological requirements for screening lipolysis effects on
release of drugs from lipid-based formulations. Although
Fed State Simulated Gastric Fluid (FeSSGF) would be in
principle more appropriate for evaluation of dispersion and
initial digestion of lipid dosage forms in the fed state, it was
not used in this study because of practical difficulties asso-
ciated with conducting lipolysis studies in this medium and
issues with filtration and analysis of solubilised drug in
FeSSGF. With respect to digestion, Jantratid et al. made the
assumption that for complete digestion of one dose lipid
formulation (corresponding to 1 ml triglycerides) approxi-
mately 1000 units of lipase would be sufficient. Thus, the
lipase content in the fed intestine exceeds by far the lipolytic
activity actually required for the digestion of lipid-based
formulations.[13] But also in the fasted intestine, it has also
been reported that the lipolytic activity required for the
digestion of the amount of lipid formulation present in a
typical soft gelatin capsule formulation is exceeded.[16–18]
Thus, the in-vivo digestion of a unit dose of lipid vehicle
should be simulated adequately by the addition of just
100 U/ml pancreatin in both media.
An important goal of this study was to design media that
not only contain sufficient lipolytic activity but also can be
used to generate results comparing lipolysis in both
prandial conditions over the duration of the lipolysis
screening test. Although this seems difficult to achieve,
given that lipolysis is controlled by many factors, including
pH, concentrations of phospholipids, bile acids and diges-
tion products, as well as the removal kinetics of the prod-
ucts of digestion, enzyme activity measurements indicate
that the pH has a particularly strong influence on lipolytic
activity.[2,10,19,20] While optimal lipase activity has been
observed at pH 7.5, a decrease of just 1 pH unit leads to a
decline of 40% in lipase activity, and lowering the pH a
further half unit to pH 6.0 leads to a further decrease of
50% in lipase activity.[2] Therefore, it was decided to set the
pH at 6.5 for both the fasted and the fed state, as this value
lies within the normal range of pH (towards the upper
619
Screening oral lipid formulations Murat Kilic and Jennifer Dressman

values observed for jejunal pH for the fed state and repre-

% solubilised danazol of the

120
senting an average value for the fasted state) and at the same
100
time provides an appropriate level of lipase activity. Conse-

applied dose
quently, early FeSSIF (rather than FeSSIF-V2 which has a 80
pH of 5.8) was selected for the fed state and FaSSIF-V2 was 60
selected for the fasted state. In the case of FaSSIF-V2, the 40
buffer capacity was increased from 10 to 25 mmol/ΔpH/L.
20
This represents a practical compromise between ensuring a
stable pH range during lipolysis on the one hand, and 0
F1 F2 F3 F4
reflecting the maximum values that have been recorded for
buffer capacity in the fasted human intestine, at maximum modified FaSSIF-V2 with pancreatin
13 mmol/ΔpH/L, on the other hand.[21] Even with this com-
Figure 3 Percentage of danazol remaining in solution after lipolysis
promise, a decrease in the pH from 6.5 to as low as pH 6.1–
screening of four lipid-based formulations using modified FaSSIF-V2
6.4, depending on the extent of digestion, was observed in with pancreatin (mean ± standard deviation, n = 3, significantly differ-
this study. Although such a decrease is not expected to ent mean values, P < 0.001).
influence the solubility of neutral APIs and would not lead
to precipitation of weak bases, it could lead to some precipi-
conditions in the simplified lipolysis screening method are
tation of a weakly acidic API if it exhibits a pKa value in this
appropriate to reproduce the results with the pH-stat and
range.
are equally able to reflect the rank order in the in-vivo
The calcium ion concentration of 5 mm is intended to
results. Thus, the hydrodynamics generated with the orbital
maintain the rate of lipolysis: during digestion fatty acids
shaker appear to provide sufficient sample dispersion. This
that are liberated from the formulation can be reacted with
is an important point because it has been recognised that
the calcium ions to form calcium soaps and thus are
hydrodynamics is one of the crucial parameters to in-vitro
removed continuously from the dosage form/medium
lipolysis.[10] Further, the incubation period of 60 min
interface. This high concentration of calcium ions (relative
appears to represent an adequate lipolysis time because the
to the physiological concentration) prevents inhibition of
extent of danazol precipitation (decrease in percentage
pancreatic lipase by fatty acids in the in-vitro test,[22]
danazol remaining in solution) is similar at this time point
whereas in vivo the concentrations of calcium and the fatty
to the results with the standard pH stat models, noting that
acids remain low because of the uptake of these across the
with the standard pH-stat set-up the lipolysis is already
gut wall – mechanisms that are not simulated in either the
nearly complete within the first 5–10 min.[23] The dilution
pH-stat or the simplified lipolysis screening set-ups.
factor of 1 : 100 (formulation : medium) used in the
In preliminary studies, a sodium oleate concentration of
lipolysis screening model is much lower than in a classical
40 mm (as in the composition of early FeSSIF suggested by
dissolution set-up (paddle or basket). Assuming a volume
Jantratid et al.) was used, but this led to excessive precipita-
of 1 ml lipid formulation in a capsule, this brings the dilu-
tion of oleate and calcium as calcium soaps. Therefore, in
tion ratio closer to a match with observed volumes of intes-
the modified version of early FeSSIF, the oleate content was
tinal fluids in vivo, which seem to be distributed as ‘water
reduced to 0.8 mm, corresponding to the concentration in
pockets’ with fluid volumes under 100 ml in each intestinal
FeSSIF-V2. According to recent studies, the fatty acid con-
segment.[25]
centration in the postprandial intestine is much closer to
this value than to the original value of 40 mm.[23,24]
Lipolysis screening for prediction of
food effects
Experimental set-up of lipolysis screening in
Figure 4 shows results from lipolysis screening under simu-
the fasted state
lated fed state conditions. The comparison of results in
Figure 3 shows results for the four danazol formulations modified early FeSSIF with modified FaSSIF-V2 enables a
using the simplified lipolysis screening method adapted for forecast of food effects. In this case example, although
fasted state conditions. The rank order of the results and the danazol still precipitated from three of the four formula-
ability to discriminate among formulations is comparable tions under simulated fed state conditions, more danazol
with the results with the pH-stat set-up obtained by Cuine remained in solution for all four formulations under condi-
and colleagues.[8] Both in-vitro methods predict the rank tions simulating the fed state than in the fasted state. This
order F1 > F2 > F3 > F4 and are commensurate with the finding is in line with the apparent danazol solubilities in
in-vivo results, where the rank order is F1 > F2 ∼ F3 > F4. modified FaSSIF-V2 without pancreatin and early FeSSIF
This preliminary assessment suggests that the choice of test without pancreatin (4.16 ± 0.34 μg/ml and 30.26 ± 2.52 μg/

620 © 2013 Royal Pharmaceutical Society, Journal of Pharmacy and Pharmacology, 66, pp. 615–623
Murat Kilic and Jennifer Dressman Screening oral lipid formulations

% solubilised danazol of the applied dose

% solubilised danazol of the

120
120
100
100
applied dose

80 80
60 60
40 40
20 20
0 0
F1 F2 F3 F4 F1 F2 F3 F4
modified early FeSSIF with pancreatin
modified early FeSSIF without pancreatin
Figure 4 Percentage of danazol remaining in solution after lipolysis modified early FeSSIF with pancreatin
screening of four lipid-based formulations using modified early FeSSIF
with pancreatin (mean ± standard deviation, n = 3, significantly differ-
ent mean values, P < 0.001). Figure 6 Comparison of danazol remaining in solution in the lipolysis
screening method using modified early FeSSIF in the presence and
absence of pancreatin (shaded bars show results without pancreatin,
with bars show results with pancreatin from Figure 4; mean ± standard
% solubilised danazol of the applied dose

120 deviation, n = 3, significantly different mean values in the presence and

absence for each formulation, P < 0.001).
100

80
60
40
20
0
F1 F2 F3
modified FaSSIF-V2 without pancreatin
modified FaSSIF-V2 with pancreatin
Figure 5 Comparison of danazol remaining in solution in the lipolysis
screening method using modified FaSSIF-V2 in the presence and
absence of pancreatin (shaded bars show results without pancreatin,
with bars show results with pancreatin from Figure 3; mean ± standard
deviation, n = 3, significantly different mean values in the presence and
absence for each formulation, P < 0.001).
ml, respectively). However, the predicted prandial differ-
ences in relative performance of the formulations cannot be
confirmed by in-vivo data because the Beagle studies in the
Cuine study were only run in the fasted state.
Evaluation of lipolysis screening results
With the simplified lipolysis screening model, it is easy to
determine the relative effects of dispersion and lipolysis on
drug precipitation, as this is merely a matter of adding or
not adding pancreatin to the test media. Results for this
comparison under simulated fasted state conditions are
shown in Figure 5.
It appears from the results in Figure 5 that the precipita-
tion of danazol is partly driven by dilution of the formula-
tion with the screening medium and partly driven by
lipolysis, with the relative influence depending on the for-
mulation type. In the case of Formulation F1, which was
composed mainly of oils (60% of the formulation con-
sisted of a 1 : 1 ratio of soya oil to maisine, it is a Type IIIa
formulation according to the LFCS) the addition of pan-
F4
creatin to the screening medium (and hence lipolysis)
resulted in 25% precipitation of danazol, whereas no pre-
cipitation was seen in the absence of pancreatin. The other
three formulations were Type IIIb or Type IV formulations
containing more than 50% surfactant. For these formula-
tions, the dilution of the formulation with the screening
medium appeared to have a greater influence on danazol
precipitation than the lipolysis. Indeed, the rank order of
results in the absence of pancreatin (i.e. where precipita-
tion is only driven by dilution of the formulation in the
medium) predicts the in-vivo study results equally well as
in the presence of pancreatin, suggesting that in the Cuine
study, the in-vivo performance was driven mainly by
dilution/dispersion effects. These results underscore the
importance of running appropriate control data before
drawing a conclusion about the reasons behind differences
in in-vivo performance of lipid formulations.
Figure 6 shows the analogous data under simulated fed
state conditions. Results from tests in early FeSSIF with
and without addition of pancreatin suggest a much differ-
ent relative impact of dilution and lipolysis on the
solubilisation performance of the four danazol formula-
tions in the fed state. For formulations F1, F2 and F3, the
drug solubilisation is not reduced at all in the medium
without lipolytic activity, that is, dilution alone does not
lead to precipitation. This is likely due to the higher

© 2013 Royal Pharmaceutical Society, Journal of Pharmacy and Pharmacology, 66, pp. 615–623 621
Screening oral lipid formulations Murat Kilic and Jennifer Dressman

concentrations of lipid components in the fed state
medium, which can solubilise the danazol. For formulation
F1, the protection against precipitation was almost com-
pletely maintained even in the presence of lipolysis. By con-
trast, over 50% of drug precipitated from formulations F2
and F3 when pancreatin was added to the screening
medium, indicating that lipolysis is the key driver of
danazol precipitation under simulated fed state conditions
for these formulations. In the case of formulation F4,
although in general terms, there was less precipitation under
simulated fed than fasted state conditions, the contribution
of lipolysis relative to dispersion also appeared to increase.
These findings demonstrate that by comparing results
run under simulated fasted conditions with those run under
simulated fed conditions (Figures 3 and 4) in the simplified
lipolysis model, in addition to attaining a qualitative under-
standing of potential food effects, a better understanding of
the relative importance of dilution and lipolysis effects on
drug precipitation can be achieved.
Conclusion
In this study, we used modified version of the intestinal
biorelevant media FaSSIF-V2 and early FeSSIF to design a
simplified screening method, which can be used to assess
the relative influence of lipolysis and dispersion on drug
solubilisation and precipitation from lipid based formula-
tions. This method includes dispersion of formulation in
medium, incubation on an orbital shaker at 37°C for 1 h
before ultracentrifugation and subsequent drug assay
of the supernatant. The method represents a simple and
efficient alternative to the standard pH-stat lipolysis model
and offers a higher throughput. By applying the screen-
ing method to previously published results for four
formulations of danazol (Pouton types III–IV), it was pos-
sible to show that the simplified screening concept was
comparable with the pH-stat set-up: the same rank order of
formulation performance was observed as in the in-vitro
pH-stat studies and was equally able to forecast the results
of the in-vivo dog study. In contrast with the standard
pH-stat models, our simplified screening method can be
used to assess dispersion and lipolysis in both the fasted and
the fed state, and therefore, it can be used to forecast food
effects on dosage form performance. Furthermore, from our
screening results, it seems to be crucial to compare formula-
tion behaviour in the presence and absence of pancreatin to
evaluate the relative contributions of lipolysis and disper-
sion to drug release and precipitation. In-vitro and in-vivo
studies of other API/lipid formulation combinations with
the simplified screening method are warranted to further
explore the applicability of this method for predicting lipid
dosage form performance before and after food intake.
Declarations
Conflict of interest
The Author(s) declare(s) that they have no conflicts of
interest to disclose.
Funding
This research received no specific grant from any funding
agency in the public, commercial or not-for-profit sectors.
Acknowledgement
This work is dedicated to the memory of Dr. Ekarat
Jantratid (1975–2010).

References emulsifying, self-emulsifying and 8. Cuine JF et al. Increasing the

1. Hauss DJ. Oral lipid-based formula-
tions. Adv Drug Deliv Rev 2007; 59:
667–676.
2. Pouton CW. Formulation of poorly
water-soluble drugs for oral
administration: physicochemical and
physiological issues and the lipid for-
mulation classification system. Eur J
Pharm Sci 2006; 29: 278–287.
3. Porter CJ et al. Enhancing intestinal
drug solubilisation using lipid-based
delivery systems. Adv Drug Deliv Rev
2008; 60: 673–691.
4. Pouton CW. Lipid formulations for
oral administration of drugs: non-
‘self-microemulsifying’ drug delivery
systems. Eur J Pharm Sci 2000; 11:
93–98.
5. Porter CJ et al. Lipids and lipid-based
formulations: optimizing the oral
delivery of lipophilic drugs. Nat Rev
Drug Discov 2007; 6: 231–248.
6. Humberstone AJ, Charman WN.
Lipid-based vehicles for the oral deliv-
ery of poorly water soluble drugs.
Adv Drug Deliv Rev 1997; 25: 103–
128.
7. MacGregor KJ et al. Influence of
lipolysis on drug absorption from the
gastro-intestinal tract. Adv Drug Deliv
Rev 1997; 25: 33–46.
proportional content of surfactant
(Cremophor EL) relative to lipid in
self-emulsifying lipid-based formula-
tions of danazol reduces oral
bioavailability in beagle dogs. Pharm
Res 2007; 24: 748–757.
9. Cuine JF et al. Evaluation of the
impact of surfactant digestion on
the bioavailability of danazol after
oral administration of lipidic
self-emulsifying formulations to
dogs. J Pharm Sci 2008; 97: 995–
1012.
10. Zangenberg NH et al. A dynamic in
vitro lipolysis model. I. Controlling
the rate of lipolysis by continuous

622 © 2013 Royal Pharmaceutical Society, Journal of Pharmacy and Pharmacology, 66, pp. 615–623
Murat Kilic and Jennifer Dressman
addition of calcium. Eur J Pharm Sci
2001; 14: 115–122.
11. Sek L et al. Characterisation and
quantification of medium chain and
long chain triglycerides and their in
vitro digestion products, by HPTLC
coupled with in situ densitometric
analysis. J Pharm Biomed Anal 2001;
25: 651–661.
12. Fatouros DG et al. Morphological
observations on a lipid-based drug
delivery system during in vitro diges-
tion. Eur J Pharm Sci 2007; 31: 85–
94.
13. Jantratid E et al. Dissolution media
simulating conditions in the proximal
human gastrointestinal tract: an
update. Pharm Res 2008; 25: 1663–
1676.
14. Glomme A et al. Comparison of a
miniaturized shake-flask solubility
method with automated potentio-
metric acid/base titrations and calcu-
lated solubilities. J Pharm Sci 2005; 94:
1–16.
© 2013 Royal Pharmaceutical Society, Journal of Pharmacy and Pharmacology, 66, pp. 615–623
15. Schamp K. Lipidartige Formulierungen
Zur Verbesserung Der Bioverfuegbarkeit
Schwer Loeslicher Arzneistoffe. Frank-
furt: JW Goethe University, 2002.
16. Keller J et al. Duodenal and ileal nutri-
ent deliveries regulate human intesti-
nal motor and pancreatic responses to
a meal. Am J Physiol 1997; 272: 632–
637.
17. Keller J, Layer P. Human pancreatic
exocrine response to nutrients in
health and disease. Gut 2005; 54: 1–28.
18. Bozkurt T et al. Effect of atropine on
intestinal phase of pancreatic secre-
tion in man. Digestion 1988; 41: 108–
115.
19. Armand M et al. Physicochemical
characteristics of emulsions during fat
digestion in human stomach and duo-
denum. Am J Physiol 1996; 271: 172–
183.
20. Lykidis A et al. Combined effect of a
lecithin and a bile salt on pancreatic
lipase activity. Comp Biochem Physiol
B Biochem Mol Biol 1997; 116: 51–55.
Screening oral lipid formulations
21. Perez de la Cruz Moreno M et al.
Characterization of fasted-state
human intestinal fluids collected from
duodenum and jejunum. J Pharm
Pharmacol 2006; 58: 1079–1089.
22. Patton JS et al. The light microscopy
of triglyceride digestion. Food
Microstruct 1985; 4: 29–41.
23. Clarysse S et al. Postprandial evolution
in composition and characteristics of
human duodenal fluids in different
nutritional states. J Pharm Sci 2009;
98: 1177–1192.
24. Vertzoni M et al. Luminal lipid phases
after administration of a triglyceride
solution of danazol in the fed state
and their contribution to the flux
of danazol across Caco-2 cell
monolayers. Mol Pharm 2012; 9:
1189–1198.
25. Schiller C et al. Intestinal fluid
volumes and transit of dosage forms
as assessed by magnetic resonance
imaging. Aliment Pharmacol Ther
2005; 22: 971–979.
623