You are on page 1of 14

Journal of Controlled Release 48 (1997) 223–236

Physicochemical characterization of lipid nanoparticles and


evaluation of their drug loading capacity and sustained release
potential
a, a b
K. Westesen *, H. Bunjes , M.H.J. Koch
a
Friedrich-Schiller-University Jena, Institute for Pharmacy, Department for Pharmaceutical Technology, Lessingstrasse 8, D-07743
Jena, Germany
b
European Molecular Biology Laboratory, Hamburg Outstation, EMBL c /o DESY, Notkestrasse 85, D-22603 Hamburg, Germany

Received 16 April 1996; revised 23 December 1996; accepted 24 December 1996

Abstract

Drug carrier systems based on lipid nanosuspensions prepared by melt emulsification present a number of severe stability
problems such as a high gelation tendency, considerable particle growth and drug expulsion. Destabilization of the
emulsified lipidic carriers is related to recrystallization of the lipids. The choice of stabilizers for colloidal lipid suspensions
is, therefore, restricted. Systematic surface modifications are thus limited. In addition, the drug payload of crystalline
nanosuspension particles is generally low. Improved stability and loading capacities were found for amorphous lipid
nanoparticles which present the characteristic signals of supercooled melts in high resolution 1 H-NMR. The NMR data
indicate that such liquid but viscous carriers can, however, not immobilize the incorporated drug molecules to the same
extent as a solid matrix. Sustained release over days or weeks as in slowly biodegraded solid matrices thus seems difficult to
achieve with a supercooled melt. Attempts to combine the advantages of the solid crystalline lipids and the amorphous nature
of the supercooled melts by generating solid but amorphous lipid suspension particles with a satisfactory long-term stability
by a variation of the lipid matrix material have hitherto not been successful. Even a satisfactory stabilization of the
a-modification using complex lipid mixtures to improve the loading capacity or to slow down the drug expulsion process
could not be achieved. The rates of the polymorphic transitions were much higher in the colloidal lipid dispersions than in
the bulk for the hard fats under investigation. Despite the fact that the properties of the lipids are superimposed with colloidal
properties, significant differences between monoacid triglycerides and complex lipids were, however, found.  1997
Elsevier Science B.V.

Keywords: Solid lipid nanoparticles; Nanoparticles of supercooled melts; Loading capacity; Molecular mobility; NMR

1. Introduction with respect to intravenous (i.v.) administration.


Decades ago submicron-sized vegetable oil-in-water
Numerous drugs are poorly soluble or unstable in (O / W) emulsions were introduced as carrier systems
aqueous media and thus cause formulation problems for poorly water-soluble drugs [1–3]. Although these
O / W emulsions are considered to be biodegradable,
*Corresponding author. Tel.: 149 3641 636781; fax: 149 3641 biocompatible and easy to manufacture, remarkably
636766. few drug-containing emulsions have reached the

0168-3659 / 97 / $17.00  1997 Elsevier Science B.V. All rights reserved


PII S0168-3659( 97 )00046-1
224 K. Westesen et al. / Journal of Controlled Release 48 (1997) 223 – 236

market as a consequence of severe formulation state. Although the raw materials chosen as matrix
problems. These problems arise from the low solu- materials are usually crystalline at room temperature,
bility of most poorly water-soluble drugs in the recrystallization does not always occur in the dis-
compositions and from the destabilizing effect of persed state [16,17]. Nanoparticles prepared from
many drugs on the emulsions. Traditional O / W trilaurin which exhibits a melting temperature around
emulsions are hardly suited for sustained release in 468C in the bulk do not recrystallize above 08C as
any case since the low viscosity of the dispersed shown by transmission electron microscopy (TEM),
liquid phase, combined with the high specific surface X-ray diffraction studies and differential scanning
area of colloidal dispersions, causes rapid drug calorimetry (DSC) [16,17]. These lipidic nanoparti-
diffusion out of the droplets [4]. In contrast, the cles are emulsion droplets of supercooled melts
mobility of drug molecules is drastically reduced in a rather than amorphous solid particles. This is in clear
solid phase which should thus make it possible to contradiction with claims from other laboratories
reduce drug leakage or hydrolysis. Drug immobiliza- [18] which are, however, not substantiated by ex-
tion in a dispersed solid lipid phase might also perimental evidence.
prevent destabilization of the colloid by the drug In recent studies it was found that trimyristin
often observed in emulsions. First attempts to de- nanoparticles can either be forced to transfer into the
velop lipid-based colloidal suspensions by replacing solid (crystalline) state under controlled conditions or
the dispersed lipid oil phase with solid triglycerides be kept in the supercooled liquid state at room
already date back to the time when the first parenter- temperature [16,17]. In the present study, trimyristin
al O / W emulsions became commercially available. dispersions were introduced as model systems to
Attempts to formulate saturated long-chain tri- study the influence of either the liquid or the
glycerides with high melting temperatures into phos- crystalline state on the loading capacity of (mono-
pholipid-stabilized aqueous suspensions by melt- acid) triglyceride nanoparticles and on the molecular
emulsification were unsuccessful because of in- mobility of drugs to derive information on the
stability, leading, for instance, to the formation of sustained release potential.
semisolid gels [5]. Nanoparticles from mixed triglycerides may be an
Since the 1980s colloidal lipid suspensions gained interesting alternative to monoacid triglyceride
renewed scientific and industrial interest as drug nanoparticles with respect to drug incorporation
carriers [6–12]. A satisfactory long-term stability of capacity because mixed triglycerides usually have a
products with lipid concentrations comparable to lower degree of crystalline order. The processing and
parenteral emulsions was, however, either not crystallization behavior as well as the structure of
achieved [8,13] or not reported. nanoparticles prepared from several hard fat sup-
In the early 1990s colloidal lipid suspensions with pository bases is, therefore, also investigated and
lipid concentrations of 10 wt% based on new stabi- compared to that of trimyristin nanoparticles.
lizer compositions and exhibiting a satisfactory long- Since immobilization of drug molecules in
term stability were described [5]. These dispersions nanoparticles of supercooled melts is not expected to
were prepared by melt-emulsification preferentially occur in the same way as in solid nanoparticles, it
using high-pressure homogenizers. The effects of was also attempted to obtain information on the
homogenization parameters and composition (matrix formation of supercooled melts in colloidally dis-
constituents, volume fraction of the dispersed phase, persed lipids, on their stability and on strategies
type and amount of emulsifying agents) on the mean preventing their occurrence. For this purpose, high
particle size of colloidal lipid suspensions turned out resolution proton nuclear magnetic resonance ( 1 H-
to be similar to those observed for O / W emulsions NMR) spectroscopy is introduced as a tool to
[14,15]. Since melt-emulsification processes are pref- distinguish between amorphous solids and liquids in
erably applied for the production of solid lipid dilute colloidal dispersions, to determine drug dis-
particles with narrow size distributions in the lower tribution within the compositions and to study the
nanometer size range, recrystallization of the molten mobility of drug molecules incorporated in the lipid
lipids must occur to obtain the desired suspension matrix.
K. Westesen et al. / Journal of Controlled Release 48 (1997) 223 – 236 225

2. Materials Table 1
Composition of drug free dispersions

Dynasan 114 (trimyristin, D 114) was provided by Matrix lipid Emulsifier blend Preparation temperature (8C)
¨ AG (Witten, Germany). The purity of the fatty
Huls D 114 PL / Tyl 70
acid fraction is approximately 95% and the hydroxyl D 114 PL / SGCa 70
value is below 5. The hard fat suppository bases D 114 PL / SGCb 70
D 114 PL 70
Witepsol H 35, H 42 and E 85, also provided by
E 85 PL / Tyl 80
¨ AG (Witten, Germany), consist of mixed
Huls E 85 PL 65
triglycerides with chain lengths mainly between 12 H 35 PL / Tyl 80
and 18 carbons. The maximum hydroxyl value of H 35 PL / SGCa 80
these commercial hard fats is 3 (H 35, H 42) or 15 H 35 PL 65
H 42 PL / Tyl 80
(E 85).
H 42 PL / SGCb 60
The following materials were from the indicated H 42 PL 65
sources and used without further purification: soy
Stabilizer composition: PL, Soy bean phospholipid Lipoid S100
bean lecithin Lipoid S100 (Lipoid KG, Ludwigs- (S100) 2.4%; PL / SGCa, S 100: 1.6%, sodium glycocholate:
hafen, Germany), sodium glycocholate (Sigma, St. 0.4%; PL / SGCb, S 100: 2.4%, sodium glycocholate: 0.6%; PL /
Louis, MO), menadione (Sigma, St. Louis, MO), Tyl, S 100: 2%, tyloxapol: 2%.
betamethasone valerate (Mainland Pharmazeutische
Fabrik GmbH, Frankfurt, Germany), ubidecarenone
was donated by Pharmacia AB (Uppsala, Sweden), temperature as the lipid melt was prepared by probe
prednisolone (Caelo, Hilden, Germany), tyloxapol sonication. This pre-mix was passed through a
USP (Sterling Organics), oxazepam (Thomae, thermostatized high-pressure homogenizer (Micron
Biberach, Germany), cortisone (Synopharm, Lab 40, APV Gaulin, Lubeck,¨ Germany) for five
¨
Barsbuttel, Germany), diazepam (Ph. Eur.), thiomer- cycles at 800 bars. The hot dispersions were filtered
sal (Synopharm, Barsbuttel,¨ Germany), retinol through 0.45 mm filters and allowed to cool down to
(Fluka, Buchs, Switzerland), glycerol 85% (Wasser- room temperature. The dispersions were split into
fuhr, Bonn, Germany), anhydrous glycerol (Merck, two fractions one of which was stored at room
Darmstadt, Germany), deuterated water (99.95%, temperature (20–258C), the other at 4–88C.
Energiteknik AB, Studsvik, Sweden), 3-(trimethyl- Particle sizes were analyzed by photon correlation
silyl)-propanesulfonic acid sodium salt (DSS) spectroscopy (PCS) with a Zetasizer 3 (Malvern
(Merck, Darmstadt, Germany), bidistilled water. Instruments) or an Autosizer 2 C (Malvern Instru-
ments) calculating the size distribution assuming
spherical particles applying exponential sampling
3. Methods without Mie correction. The mean particle size was
calculated from the number distribution as the diam-
The lipids were heated to at least 108C above their eter of the equivalent hydrodynamic sphere.
melting point. The phospholipid was dispersed in the Thermal analysis was performed in a Perkin Elmer
melt by probe sonication (Bandelin Desintegrator DSC 2-C calorimeter using accurately weighed
HD 200, Bandelin Electronics, Berlin, Germany) samples of 1–2 mg for bulk material or 10–20 mg
until the dispersion appeared clear. For the drug for dispersions. Heating curves were recorded with a
incorporation study, the drug was dissolved in the scan rate of 5 or 108C / min. Crystallization data was
triglyceride-phospholipid mixture. The composition obtained from thermally untreated samples or after
of the dispersions are given in Table 1 and Table 2. keeping the samples around 208C above their melting
The co-emulsifier was dissolved in the aqueous point for at least 10 min. The cooling rate was
phase containing 0.01% thiomersal as a preservative 58C / min. The melting and recrystallization tempera-
if required. Glycerol at 2.25% was added for isotoni- tures given correspond to the maxima or minima,
zation. A predispersion of the lipid in this aqueous respectively, in the DSC curves. Polymorphic forms
phase previously heated to approximately the same were assigned by correlation with the X-ray data.
226 K. Westesen et al. / Journal of Controlled Release 48 (1997) 223 – 236

Table 2
Composition and stability of drug loaded dispersions
Matrix Drug Drug Solubility Emulsifier Phase separation
lipid concentration a in lipid blend of crystalline
(w%) melt drug (suspension)
b
D 114 Oxazepam 5 *
D 114 Diazepam 5 1 PL / SGCb *
D 114 Diazepam 3 1 PL / SGCb *
b
D 114 Cortisone 5 *
D 114 Betamethasone 5 1 PL / SGCb 1 Drug crystals also
valerate observed in emulsion
b
D 114 Prednisolone 5 *
D 114 Retinol 5 1 PL / SGCb *
D 114 Menadione 2 1 PL / Tyl 1 c Macroscopic drug
crystals after 2 days
of cold storage
D 114 Menadione 1 1 PL / Tyl *
E 85 Menadione 2 1 PL / Tyl 1 c Macroscopic drug
crystals after 4 days
of cold storage
H 42 Menadione 2 1 PL / Tyl 1 c Macroscopic drug
crystals after 6 days
of cold storage
D 114 Ubidecarenone 50 1 PL / SGCb *
D 114 Ubidecarenone 10 1 PL / SGCb *
D 114 Ubidecarenone 5 1 PL / SGCb *
a
Drug concentration is related to the lipid phase (matrix lipid1drug).
b
No dispersions were prepared if the incorporated drug was insoluble in the lipid melt.
c
Macroscopic and microscopic investigations were performed regularly within several days after preparation.
All dispersions were prepared at 708C applying five homogenization cycles at 800 bar.

X-ray diffraction measurements were performed samples. 908 pulse widths were used with a pulse
on the double focusing monochromator mirror ca- repetition time of 14 s allowing complete relaxation
mera X33 of the EMBL in HASYLAB on the of the system. The relaxation time T 1 was deter-
storage ring DORIS of the Deutsches Elektronen mined by the inversion recovery method to be 3.2 s
Synchrotron (DESY) at Hamburg as described previ- for the trimethylsilyl signal of DSS. FIDs were
ously [17]. Fourier transformed without line broadening.
High resolution proton nuclear magnetic resonance The drug loaded dispersions were examined in a
( 1 H-NMR) spectra of samples prepared in deuterium Zeiss Laboval 4 microscope (Zeiss, Oberkochen,
oxide were obtained on a WM-400 (drug-free disper- Germany) equipped with crossed polarizers and a
sion) and a DRX-400 (drug-loaded dispersions) l-sheet. The investigations were performed after 1
¨
NMR spectrometer (Bruker, Rheinstatten, Germany) day of storage if not otherwise specified.
operating at 400 MHz and ambient temperature.
Drug-containing dispersions were investigated with-
out adding an internal reference using 308 pulse 4. Results and discussion
widths. In case of the drug-free D 114 nanodisper-
sions, 3-(trimethylsilyl)-propanesulfonic acid sodium After melt-emulsification the dispersions were
salt (DSS) (d50.00 ppm) was used as an internal split into two fractions one of which was stored at
reference for quantitative analysis. DSS was dis- room temperature (20–258C), the other one at 4–
solved in D 2 O (about 25 mg DSS accurately weighed 88C. If the dispersions were stabilized with either the
in 10 ml D 2 O) and added in 1:1 portions to the phospholipid / tyloxapol blend or with the phos-
K. Westesen et al. / Journal of Controlled Release 48 (1997) 223 – 236 227

pholipid / bile salt blend both fractions exhibit ap- enriched in phase separated drug (Table 2). The
proximated mean particle sizes between 60 and 150 detection of these drug crystals in the milky disper-
nm and are thus in the lower colloidal size range. sions is often difficult. The use of X-ray diffraction
Stabilization with the phospholipid / tyloxapol blend is, in most cases, not possible due to problems in the
usually yields smaller particles than with the phos- preparation of homogenous samples (sedimentation
pholipid / bile salt blend. The particle sizes in the of drug crystals) and detection limits of the method.
drug-containing dispersions are comparable to those Detection and quantification of phase separated drug
of the corresponding drug-free dispersions. by DSC is usually also impossible because the high
Dispersions stabilized with the phospholipid only water content of the dispersions limits the maximum
deviate in their behavior. The purely phospholipid temperature of the heating runs. Moreover, the first
stabilized dispersions of E 85 and H 42 kept at room order transition endotherms of the matrix materials
temperature form highly viscous or soft gel-like are extremely broadened for the colloidally dispersed
systems within one day of storage. More rigid gels materials stabilized with emulsifier blends. This
are formed at refrigerator temperature. D 114 and H holds especially true for the complex lipids (Fig. 8).
35 dispersions remain liquid at room temperature. In contrast, polarizing light microscopy is an appro-
Whereas the D 114 dispersion also turns rapidly into priate technique for the detection of expelled, crys-
a semisolid gel at 58C the H 35 dispersions remains tal-forming drugs. The crystals of the model drugs
liquid even at this low temperature at least for can be identified in the dispersions via their optical
several days. Despite the fact that there are obvious anisotropy and simple geometric shape. In some
similarities between lipid nanosuspensions and col- cases, the drug crystals can also be detected by
loidal oil-in-water (o / w) emulsions with respect to macroscopic observation of the dispersions. The
the preparation method and the chemical composi- phase separation of incorporated drug may occur
tion, all suspension formulations under investigation after various time intervals depending on the com-
stabilized with phospholipids only exhibit a pro- position of the matrix material. For menadione-
nounced tendency to form semisolid, water-rich gels loaded dispersions, the phase separation is considera-
especially at low temperatures. The critical tempera- bly retarded in systems based on complex glycerides.
ture depends on the matrix material. The gel forma- At this stage comprehensive physicochemical
tion indicates clear differences in the properties of characterization of the colloidal dispersions seemed
colloidal lipid emulsions and suspensions but can be essential to understand the phenomena and to check
circumvented by the addition of a co-surfactant. the conclusions drawn from bulk properties of lipids.
To evaluate the loading capacity and the sustained In order to be able to generalize the results and to
release potential of lipid nanoparticle dispersions correlate the physical state of the lipids in the
based on monoacid triglycerides or complex dispersions with the drug loading capacity and their
glyceride mixtures, a selection of poorly water sustained release potential, initial studies were per-
soluble model drugs were added in various con- formed on the drug free compositions.
centrations to the basic compositions. The composi- None of the drug free dispersions stabilized with
tion of the drug-loaded systems and the observed one of the stabilizer blends stored at room tempera-
instability phenomena are given in Table 2. ture for more than three months displays any small
Ubidecarenone is the only drug that can be added to or wide angle reflection in their X-ray patterns (Fig.
the basic compositions even in high concentrations 1). They also do not display any thermal event
without destabilizing the systems. Several model (melting endotherms or glass transitions) in DSC
drugs cannot even be dissolved in the lipid melts in heating runs (Fig. 2). Whereas X-ray diffraction data
concentrations as low as 5 wt% related to the lipid only allow to differentiate between crystalline and
— demonstrating the low solubilization capacity of amorphous material, DSC data can be used to
the molten lipids for foreign poorly water-soluble differentiate between amorphous solids and liquids.
compounds already in the heat. Many of the few The absence of a glass transition in the DSC
successfully drug-loaded dispersions exhibit a poor thermograms hints at a liquid state of the dispersed
storage stability resulting in the formation of crystals lipid phase although it cannot be excluded in case of
228 K. Westesen et al. / Journal of Controlled Release 48 (1997) 223 – 236

ture display strong signals which can be assigned to


the glycerides (Fig. 3) indicating the high mobility of
these molecules characteristic of the liquid state.
This result confirms earlier assumptions [16,17] that
the colloidal dispersion kept at room temperature is
an emulsion of a supercooled melt. To our knowl-
edge, this novel type of colloidal drug carrier has not
been described by others so far.
In contrast, cooling the freshly melt-emulsified
dispersions to refrigerator temperature results in
dispersions that display X-ray reflections in the small
and wide angle regions pointing to the crystalline
nature of the particles (Fig. 4). These dispersions
also exhibit melting endotherms in DSC heating
runs. DSC and X-ray diffraction data indicate that
Fig. 1. Small angle X-ray scattering patterns of phospholipid / solidification of all colloidally dispersed lipids under
tyloxapol stabilized nanoparticle dispersions stored at room tem-
investigation results in crystalline material. The
perature for more than 4 months (10 days in case of H 35) (s52
sin Q / l, 2 Q: scattering angle). The curves have been displaced corresponding 1 H-NMR spectrum of the D 114
vertically for better visualization. dispersion is dominated by signals which can be
assigned to the stabilizers including glycerol (Fig. 3).
This is indicative of a highly restricted mobility of
aqueous colloidal dispersions of amorphous material, the glyceride molecules typical of the solid state.
that a glass transition would occur below the de- Obviously, cooling to refrigerator temperatures in-
tection limit. Additional investigations on the phys- duces solidification of the glycerides. Since the
ical state of the submicron-sized D 114 were, formation of crystalline material upon cooling was
therefore, performed by 1 H-NMR spectroscopy. also observed in earlier studies for a series of melt-
Spectra taken after 1 day of storage at room tempera- emulsified monoacid triglycerides [16,17], the results
may be considered in more general terms indicating
that solidification of colloidally dispersed lipids
commonly results in crystalline material and that the
chance to obtain an amorphous but solid lipid matrix
via supercooling of colloidally dispersed lipid melts
seems low. The line broadening of the reference
signal (d50 ppm) in the NMR spectrum of the D
114 suspension (Fig. 3) may result from the surface
activity of DSS. Adsorption of DSS molecules on
solid particles might lead to a more restricted
molecular mobility than adsorption at the liquid-
liquid interface of emulsion droplets. The phenom-
enon is currently under investigation.
The crystallization temperature of H 35 nanoparti-
cles is so low that solidification does not always
occur, not even upon cold storage. Therefore, sys-
tematic DSC studies were performed to determine
the crystallization temperature of the supercooled
Fig. 2. DSC heating curves of phospholipid / tyloxapol stabilized
nanoparticle dispersions stored at room temperature for more than particles (Fig. 5). The crystallization temperature
3 months. The curves have been displaced along the ordinate for decreases in the order E 85, H 42, D 114, H 35. This
better visualization. is in accordance with the macroscopic observations
K. Westesen et al. / Journal of Controlled Release 48 (1997) 223 – 236 229

Fig. 3. 1 H-NMR spectra of a drug free trimyristin dispersion (stabilized with 2.4% S 100 and 0.6% sodium glycocholate) stored 1 day at
208C (left) or 58C (right).

on the gel formation tendencies of the various matrix nanosuspensions. Comparison between the melting
materials and explains the deviating stabilities on temperature of the cold stored nanoparticles and the
either room or refrigerator temperature of the differ- crystallization temperature (Fig. 5) reveals a pro-
ent compositions if stabilized with phospholipids nounced tendency towards supercooling. The differ-
only. The stabilization requirements of emulsions of
supercooled melts are obviously more similar to
those of conventional o / w emulsions than to those of

Fig. 5. DSC melting and crystallization temperatures of nanoparti-


Fig. 4. Small (SAXS) and wide (WAXS) angle X-ray scattering cle dispersions stabilized with the phospholipid / tyloxapol blend.
curves of phospholipid / tyloxapol stabilized dispersions cold The melting temperatures were obtained from samples cold stored
stored for about 1 month (3 months in case of H 35). The curves for at least several days (H 35 nanoparticles had been previously
have been displaced vertically for better visualization. crystallized at 08C).
230 K. Westesen et al. / Journal of Controlled Release 48 (1997) 223 – 236

ence between melting and crystallization temperature


is, however, distinctly smaller for dispersed hard fats
than for trimyristin or other monoacid triglycerides
[16,17]. This phenomenon may be attributed to the
fact that longer chain components in the complex
hard fat mixtures may serve as nucleation centers.
Hard fat bulk materials display two DSC ex-
otherms upon cooling of the melts whereas hard fat
nanoparticles crystallize in a single thermal event at
lower temperatures than the bulk (Fig. 6). The
processes during crystallization of the particles must
thus be different from those in bulk material. X-ray
crystallization studies on E 85 indicate that the
polymorphic transition into the more stable b9-poly-
morph occurs faster in the bulk than in nanoparticles
(Fig. 7). This phenomenon may among others be
attributed to the higher crystallization temperature of
the bulk material.
The course of polymorphic transitions is distinctly
different in hard fat nanoparticles than in trimyristin.
In contrast to the trimyristin dispersion freshly
crystallized hard fat particles tend to display several
diffuse melting maxima in the DSC (Fig. 8). The
melting endotherm with the highest melting point is
the one that remains after storage. This behavior
points to slow polymorphic transitions within the
particles. Time-resolved X-ray crystallization studies

Fig. 7. X-ray crystallization (cooling rate 0.31258C / min) studies


Fig. 6. DSC cooling curves of Witepsol E 85 melts in the bulk and of Witepsol E 85 in the bulk (top), in colloidal dispersion (middle)
in colloidal dispersion (stabilized with the phospholipid / tyloxapol and colloidal D 114 (bottom). The colloidal dispersions were
blend). The curves have been displaced vertically for better stabilized with the phospholipid / tyloxapol blend. The curves have
visualization. been displaced vertically for better visualization.
K. Westesen et al. / Journal of Controlled Release 48 (1997) 223 – 236 231

mation on the structure of the nanoparticles, e.g.


from X-ray diffraction, is thus questionable.
For hard fat suppository bases in the bulk, an
intermediate b-form has been described as the
‘stable’ polymorph [21]. The wide angle reflections
of this intermediate b-form were also found in an
earlier study on dispersions of the hard fat Witepsol
W 35 [19]. In contrast, none of the hard fat
nanoparticle dispersions in the present study display
the strong reflection at 0.42 nm which is characteris-
tic of this intermediate b-form. Instead, the reflec-
tions arising from colloidally dispersed E 85, H 35
and H 42 resemble those of the regular b-form of
triglycerides. The comparison of the behavior of
Witepsol W 35 with that of the three hard fats under
Fig. 8. DSC heating curves of freshly crystallized dispersions investigation in the present study indicates that even
stabilized with the phospholipid / tyloxapol blend. The curves have within the group of hard fats there is no consistent
been displaced vertically for better visualization.
behavior. Nevertheless, certain conclusions can be
drawn taking the chemical composition of the four
confirm that the transition rates of hard fat nanoparti- complex lipids into account. The three hard fats of
cles are lower than those of trimyristin dispersions the present study contain only comparatively small
(Fig. 7). These differences may be due to the more amounts of partial glycerides and differ mainly with
complex structure of the mixed fats impeding order- respect to the fatty acid pattern. In contrast, Witepsol
ing of the molecules. W 35 (hydroxyl value between 40 and 50) is a
Compared to their bulk material, however, the mixture of mono-, di- and triglycerides therefore
hard fat nanoparticles seem to transform more rapid- differing also significantly in physicochemical prop-
ly into the stable b-polymorph. While the thermally erties from the other three.
unstressed hard fat raw materials under investigation The X-ray long-spacings of hard fat nanoparticles
are in the metastable b9-form the solidified dispersed are larger than those of the corresponding bulk
materials display wide angle X-ray reflection patterns materials although they are due to a more stable
similar to those of the stable b-form of monoacid polymorph (Fig. 9). This may result from lattice
triglycerides after one or three months of storage
(Fig. 4). The faster polymorphic transitions in lipid
nanoparticles have been attributed to the small size
of the crystallites [20]. Compared to those of the
trimyristin dispersion the X-ray reflections of the
hard fat dispersions are less intense, especially for H
35 which may be due to lattice imperfections or
incomplete crystallization of the dispersed hard fats.
Due to the different polymorphic forms of the hard
fat nanoparticles and their bulk material it is not
possible to quantify the crystallinity of these
nanoparticles by comparison of their melting en-
thalpy with that of the bulk (DSC recrystallization
index according to [20]) since different transition
enthalpies are expected for the different polymorphs. Fig. 9. X-ray long-spacings of the bulk materials compared to
Interpretation of DSC data for colloidally dispersed those of the phospholipid / tyloxapol stabilized nanosuspensions
complex glyceride mixtures without further infor- cold stored for at least 1 month.
232 K. Westesen et al. / Journal of Controlled Release 48 (1997) 223 – 236

imperfections in the region of the methyl end group than in dispersed monoacid triglycerides but colloi-
planes or to a different chain tilt in bulk material and dal dispersions of supercooled hard fats still display
nanoparticles. Defects in the crystal lattice may also a considerable storage stability.
contribute to the depression of the DSC melting The cold stored solidified hard fat nanoparticles
temperature in these hard fat nanoparticles (b-modi- investigated here transform into a more stable poly-
fication) compared to their bulk material (b9-modi- morph than their bulk material within a comparative-
fication) which is much larger in hard fats than in ly short time of storage. Nevertheless, their structure
nanoparticles from simple triglycerides, e.g. trimyris- appears to be less ordered than in bulk material
tin (Fig. 10). The small size (below 100 nm) of the which should be favorable with respect to the drug
phospholipid / tyloxapol stabilized hard fat nanoparti- incorporation capacity. All colloidal hard fat disper-
cles may be an additional reason for their low sions have, however, to be considered as highly
melting temperature. dynamic systems which complicates an early evalua-
The comparison of the physicochemical properties tion of the final drug loading capacities. The chances
of the bulk and the colloidally dispersed lipids for efficient incorporation of foreign compounds is
indicate that fundamental differences exist which known to decrease with increasing crystal lattice
make it impossible to derive the properties of order. Annealing of the crystal lattice and poly-
dispersions from those of the bulk material in a morphic transitions can, therefore, result in sustained
simple way. Colloidal dispersions of supercooled expulsion of the drug upon storage as demonstrated
melts are not only formed by monoacid triglycerides for menadione (Table 2). Such retarded destabilizing
such as trimyristin or trilaurin [16,17] but also by processes can almost be excluded for the rapidly
complex lipid mixtures and represent an unexpected transforming monoacid triglyceride trimyristin
novel type of colloidal lipid drug carrier systems. (Table 2) and are typically not observed for tradi-
The crystallization and melting behavior of solid tional emulsions. Nanoemulsions of supercooled
nanoparticles prepared from hard fat suppository melts are, therefore, less critical in this respect than
masses is, however, different from those of dispersed nanosuspensions.
trimyristin resulting among others in a different time The drug loading characteristics observed for the
course of drug expulsion (Table 2). Supercooling model drugs may be explained on the basis of the
effects are less pronounced in hard fat dispersions physicochemical data for the colloidally dispersed
lipids. Several of the successfully drug-loaded dis-
persions exhibit a poor storage stability. Colloidal D
114 particles exhibit a higher loading capacity for
menadione if kept at room temperature i.e. in the
liquid state of a supercooled melt. Solidification of D
114 results in crystallization and drug expulsion
indicating that the drug is mainly located in the lipid
phase of the emulsions. The time course of phase
separation from a solid lipid carrier is, however,
dependent on the composition of the matrix material
since drug expulsion occurrs after different time
intervals when different types of matrices (D 114, E
85, H 42) are used (Table 2). This behavior is related
to the different time course of polymorphic transi-
tions of the matrix lipids. Crystallization in the a-
modification and slow polymorphic transitions result
in a delayed expulsion of drug from the hard fat
Fig. 10. DSC melting temperatures of the lipid in the bulk and in
colloidal dispersions (stabilized with the phospolipid / tyloxapol nanoparticles. Another reason may be different an-
blend) cold stored for at least several days (H 35 nanoparticles nealing properties of the lipids.
had been previously crystallized at 08C). High resolution NMR spectroscopy was applied as
K. Westesen et al. / Journal of Controlled Release 48 (1997) 223 – 236 233

a non-destructive method to the native drug-loaded distributes preferentially towards the liquid lipid
systems since neither X-ray diffraction nor DSC phase in this two-phase system. The high molecular
experiments allow a localization of the drug in mobility observed for all drugs under investigation
colloidal dispersions among others due to detection within the liquid lipid phase indicates that such an
limits (see above). The 1 H-NMR spectra on the emulsion of a supercooled melt cannot be used to
colloidal drug free D 114 dispersions illustrate that obtain a sustained release of drugs over days or
significant changes in molecular mobility due to weeks under sink conditions.
phase transitions can be detected (Fig. 3). The 1 H- Solidification of D 114 not only results in a loss of
NMR spectra of colloidal D 114 dispersions loaded the strong triglyceride signals but also in a consider-
with menadione (1 wt% related to the dispersed able broadening or loss of the drug signals as
phase), diazepam (3 wt%), and ubidecarenone (5, 10 demonstrated for diazepam (Fig. 11). The mobility
or 50 wt%) were also recorded. These nanodisper- of the drug molecules in the nanosuspensions is
sions were studied by NMR in the state of a drastically reduced indicating that these drugs are
supercooled melt and after solidification of the lipid. associated with D 114 since no other structure in the
The detection of drug signals in the spectra is, of systems is observed to transfer irreversibly into a
course, easier for high drug loads and for drugs much more rigid state on cooling to refrigerator
exhibiting signals clearly separated from those of the temperature despite final storage of the dispersion at
carrier dispersion. The drugs have a comparatively room temperature. Even the fairly water soluble
high molecular mobility in the emulsions of super- diazepam is not expelled from the recrystallizing
cooled D 114 as demonstrated for diazepam (Fig. lipid matrix into the water phase. The NMR data
11). The considerable line widths of the drug signals indicate that such a glyceride suspension may serve
indicate, however, that the signals result from mole- as a real sustained release system and that the model
cules which are not dissolved in the aqueous phase drugs may be released more slowly from the D 114
but in a viscous phase, i.e. mainly in the dispersed nanosuspensions than from the D 114 nanoemulsions
lipid phase. This is especially interesting in the case of supercooled melts.
of diazepam which has a solubility in aqueous media In contrast to all other poorly water soluble drug
around 0.3 wt% [22], i.e. in the concentration range substances tested so far, ubidecarenone is the only
of the loading experiment. Diazepam obviously one that can be added to the basic composition even

Fig. 11. 1 H-NMR spectra of a trimyristin emulsion (left) and suspension (right) containing 3% diazepam related to the dispersed phase. The
dispersions were stabilized with 2.4% S 100 and 0.6% sodium glycocholate. The arrows indicate the position of the drug signals.
234 K. Westesen et al. / Journal of Controlled Release 48 (1997) 223 – 236

in high concentrations without any destabilization of that a fraction of the drug is strongly bound to the
the system. Its specific feature is reflected in the solidified triglyceride and that ubidecarenone is
NMR spectra of D 114 nanodispersions loaded with distributed between two different locations in the
5, 10 or even 50 wt% ubidecarenone (Fig. 12). The colloidal system. The intensity of the remaining drug
shape of the ubidecarenone signals in the spectra of signals in the ubidecarenone loaded nanosuspensions
the nanoemulsions indicates a similarly restricted increases with the drug load of the system. The line
drug mobility as in the other drug loaded D 114 widths of the signals indicates that the drug mole-
nanoemulsions suggesting that no significant cules still have a mobility similar to that in the
amounts of the drug are dissolved in the continuous nanoemulsion. In contrast to the dispersions loaded
phase. The spectra of the nanosuspensions containing with 5 and 10% ubidecarenone, signals of liquid
the high drug loads of either 10 or 50% still contain triglyceride are detected in the spectrum of the cold
signals of ubidecarenone. The intensity of the signals stored dispersion containing 50% ubidecarenone in
is, however, reduced in comparison with the corre- the lipid phase suggesting that not all triglyceride has
sponding drug signals in the emulsions suggesting crystallized in this dispersion due to eutectic be-
havior with ubidecarenone. This assumption is sup-
ported by DSC measurements indicating that drugs
may alter the physicochemical properties of lipid
nanoparticles dramatically.

5. Conclusions

Important physicochemical properties of lipids in


the bulk and in the colloidal state differ significantly.
Physicochemical characterization demonstrated that
colloidal drug carriers based on melt-emulsified solid
glycerides are complex systems due to their crys-
tallization behavior, the rate of polymorphic transi-
tions and the potential formation of supercooled
melts. The formation of emulsified supercooled melts
which are stable on storage depends on the con-
ditions chosen for the cooling step of the hot melt-in-
water nanoemulsions. Nanoemulsions of supercooled
melts which deviate substantially from solid lipid
nanoparticles in their properties represent a new
alternative drug carrier system.
The drug loading capacity of both types of colloi-
dal lipidic carriers is limited owing to the generally
low solubilization capacity of the molten lipids for
many poorly water-soluble drugs. The drug-loading
capacity of nanoparticles representing supercooled
melts was shown to be higher than that of crys-
tallized nanoparticles. Crystallization of the dis-
persed glycerides can lead to phase separation of
Fig. 12. 1 H-NMR spectra of trimyristin dispersions stored at incorporated drugs. Neither amorphous solids nor
20–258C (left column) and 4–88C (right column) containing 5%
(top), 10% (middle) and 50% (bottom) ubidecarenone related to any sufficiently stabilized a -modification could be
the dispersed phase. All dispersions were stabilized with 2.4% S obtained by variations of the lipid matrices. Complex
100 and 0.6% sodium glycocholate. glyceride mixtures such as hard fats may, however,
K. Westesen et al. / Journal of Controlled Release 48 (1997) 223 – 236 235

possess a higher drug-loading capacity in the crys- [2] L.C. Collins-Gold, R.T. Lyons and L.C. Bartholow, Parenter-
talline state due to their lower crystallinity as com- al emulsions for drug delivery. Adv. Drug Deliv. Rev. 5
(1990) 189–208.
pared to pure monoacid triglycerides. Nanodisper-
[3] R.J. Prankerd and V.J. Stella, The use of oil-in-water
sions of complex glycerides represent highly dy- emulsions as a vehicle for parenteral drug administration. J.
namic systems on storage which complicates an early Parent. Sci. Techol. 44 (1990) 139–149.
evaluation of the final drug loading capacity. [4] B. Magenheim, M.Y. Levy and S. Benita, A new in vitro
All poorly water-soluble drugs investigated in technique for evaluation of drug release profile from colloi-
NMR studies were distributed towards the lipid dal carriers—ultrafiltration technique at low pressure. Int. J.
phase in emulsions of supercooled melts. Drug Pharm. 94 (1993) 115–123.
[5] B. Siekmann and K. Westesen, Submicron-sized parenteral
molecules incorporated in particles of supercooled carrier systems based on solid lipids. Pharm. Pharmacol.
melts have a high mobility making these systems Lett. 1 (1992) 123–126.
unsuitable for sustained release over days, weeks or ¨
[6] P. Speiser, Lipidnanopellets als Tragersystem ¨ Arzneimit-
fur
months. Drug mobility is drastically reduced in tel zur peroralen Anwendung, European Patent Application
solidified (crystallized) nanoparticles. If crystalliza- EP 0 167 825, January 15, 1986.
tion does not lead to drug expulsion from the crystal [7] M.R. Gasco and S. Morel, Lipospheres from microemul-
sions, Il Farmaco 45 (1990) 1127–1128.
lattice, these carriers may have a good potential as
[8] A.J. Domb and M. Maniar, Lipospheres for controlled
sustained release systems. delivery of substances, PCT Patent application WO 91 /
Trimyristin nanoparticles provide an excellent 07171, May 30, 1991.
model system to study the effects of supercooling ¨ ¨ and B. Bergenstahl,
[9] B. Sjostrom ˚ Preparation of submicron
and crystallization on the molecular mobility of drug particles in lecithin-stabilized o / w emulsions. I. Model
drugs incorporated in melt-emulsified glyceride studies of the precipitation of cholesteryl acetate. Int. J.
nanoparticles as well as on stability of these systems. Pharm. 88 (1992) 53–62.
[10] M.R. Gasco, Method for producing lipid microspheres
At a more general experimental level it was shown
having a narrow size distribution, US Pat. No. 5,250,236,
that high resolution proton NMR is a powerful tool October 10, 1993.
to distinguish between amorphous solids and liquids ¨
[11] R.H. Muller ¨
and J.S. Lucks, Arzneistofftrager aus festen
in dilute colloidal systems, to determine drug dis- Lipidteilchen — Feste Lipid Nanospharen, ¨ German Patent
tribution within the systems and to study the mobility Application DE 41 31 562, March 25, 1993.
of drug molecules incorporated in the lipid [12] K. Westesen and B. Siekmann, Solid Lipid Particles, Par-
ticles of Bioactive Agents and Methods for the Manufacture
nanoparticles. and Use Thereof, PCT Patent Application WO 94 / 20072,
September 15, 1994.
[13] P. Kecht-Wyrsch, Hochdisperse Glycerid-Mikropartikel als
Acknowledgments ¨
perorales Arzneitragersystem, ¨
Ph.D. thesis, Eidgenossische
¨
Technische Hochschule Zurich, 1987.
[14] B. Siekmann, Untersuchungen zur Herstellung und zum
¨ AG, D-Witten, and Lipoid
The authors thank Huls
Rekristallisationsverhalten schmelzemulgierter intravenos ¨ ap-
KG, D-Ludwigshafen, for supply of raw materials plizierbarer Glyceridnanopartikel, Ph.D. thesis, University of
and Pharmacia Oncology Immunology AB, S-Hel- Braunschweig, 1994.
singborg for financial support. We would like to [15] B. Siekmann and K. Westesen, Melt-homogenized solid lipid
thank Dr. V. Wray, C. Kakoschke and B. Jaschok- nanoparticles stabilized by the nonionic surfactant tyloxapol.
Kentner, National Research Center for Biotechnol- I. Preparation and particle size determination. Pharm. Phar-
macol. Lett. 3 (1994) 194–197.
¨
ogy, Braunschweig, Germany, and Dr. W. Gunther, [16] K. Westesen and H. Bunjes, Do nanoparticles prepared from
Institute for Organic Chemistry and Macromolecular lipids solid at room temperature always possess a solid lipid
Chemistry, Jena, Germany, for support in the 1 H- matrix?, Int. J. Pharm. 115 (1995) 129–131.
NMR experiments. [17] H. Bunjes, K. Westesen and M.H.J. Koch, Recrystallization
tendency and polymorphic transitions in triglyceride
nanoparticles. Int. J. Pharm. 129 (1996) 159–173.
[18] C. Schwarz, W. Mehnert, J.S. Lucks and R.H. Muller, ¨ Solid
References lipid nanoparticles (SLN) for controlled drug delivery. J.
Control. Release 30 (1994) 83–96.
[1] A. Wretlind, Development of fat emulsions. J. Parent. Enteral [19] K. Westesen, B. Siekmann and M.H.J. Koch, Investigations
Nutr. 5 (1981) 230–235. on the physical state of lipid nanoparticles by synchrotron
236 K. Westesen et al. / Journal of Controlled Release 48 (1997) 223 – 236

radiation X-ray diffraction. Int. J. Pharm. 93 (1993) 189– [22] Pharmazeutische Stoffliste (List of Pharmaceutical Sub-
199. ¨
stances), Arzneiburo der Bundesvereinigung Deutscher
[20] B. Siekmann and K. Westesen, Thermoanalysis of the ¨
Apothekerverbande (ABDA), Werbe- und Vertrieb-
recrystallization process of melt-homogenized glyceride sgesellschaft Deutscher Apotheker m. b. H., Frankfurt am
nanoparticles. Colloids Surf. B 3 (1994) 159–175. Main, 1993.
¨
[21] K. Thoma, P. Serno and D. Precht, Rontgendiffraktometris-
cher Nachweis der Polymorphie von Harfett. Pharm. Ind. 45
(1983) 420–425.

You might also like