Cyclodextrins as Solubilizers:
Formation of Complex Aggregates
PHATSAWEE JANSOOK, SERGEY V. KURKOV, THORSTEINN LOFTSSON
Faculty of Pharmaceutical Sciences, University of Iceland, Hofsvallagata 53, IS-107 Reykjavik, Iceland
Received 5 March 2009; revised 19 May 2009; accepted 20 May 2009
Published online 7 August 2009 in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/jps.21861
ABSTRACT: Three different techniques were applied to investigate aggregation of drug/
cyclodextrin complexes, that is, drug permeation through semi-permeable membranes,
determination of changes in the value of activity coefficients of drug/cyclodextrin
complex solutions and transmission electron microscopy (TEM). The aqueous solutions
studied contained g-cyclodextrin, 2-hydroxypropyl-g-cyclodextrin or mixtures thereof,
and hydrocortisone, amphotericin B, diclofenac sodium or indomethacin. The permea-
tion studies indicated that drug/cyclodextrin complex monomers (i.e., unaggregated
complexes) were dominating at cyclodextrin concentrations below 5% (w/v). Then
formation of aggregates gradually increased with increasing cyclodextrin concentration
until all increase in dissolved drug/cyclodextrin complexes was due to formation of
cyclodextrin aggregates. This happened even though the observed phase-solubility
diagrams were linear, that is, were of AL-type. The activity coefficients showed positive
deviation from ideal state. This positive deviation is due to concurrent of several
processes, that is, hydration, aggregation and complex formation. The observed devia-
tions from ideality indicated that complex aggregates were formed in the aqueous
complexation media. TEM pictures showed formation of aggregates in both pure
aqueous cyclodextrin solutions as well as in cyclodextrin solutions that had been
saturated with hydrocortisone. The aggregate diameter was between 10 and 100 nm
but larger aggregates with diameter of about 200 nm were formed through assemble of
smaller aggregates. ß 2009 Wiley-Liss, Inc. and the American Pharmacists Association J Pharm
Sci 99:719 729, 2010
Keywords: cyclodextrin; complex; aggregate; permeation; activity coefficient; TEM
INTRODUCTION in vitro and in vivo, and cyclodextrins can be found
in just over 30 marketed drug products world-
Cyclodextrins are hydrophilic cyclic oligosac- wide.1 In aqueous solutions free drug and
charides with a lipophilic central cavity. In cyclodextrin molecules are in dynamic equili-
aqueous solutions cyclodextrins are able to brium with the complexes and every complex
solubilize many hydrophobic drugs by taking up unity is generally assumed to be free and
some lipophilic moiety of the drugs into the cavity, independent of other complexes as well as of
that is, through formation of water-soluble inclu- other excipients found in the solution.2 It is
sion complexes. Aqueous cyclodextrin solutions however becoming increasingly apparent that
are frequently used as nontoxic solvents during such assumptions may not be universally applic-
screening of new chemical entities (NCE) both able or all encompassing. Specifically, there is a
growing body of evidence that supports the
important contribution of noninclusion aspects
Correspondence to: Thorsteinn Loftsson (Telephone: þ354-
for drug solubilization by cyclodextrins including
525-4464; Fax: þ354-525-4071; E-mail: [e-mail])
surfactant-like effects and molecular aggregation,
Journal of Pharmaceutical Sciences, Vol. 99, 719 729 (2010)
ß 2009 Wiley-Liss, Inc. and the American Pharmacists Association that is, formation of particulate systems.3 10 It is
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 2, FEBUARY 2010 719
720 JANSOOK, KURKOV, AND LOFTSSON
known that spontaneous opalescence occurs in diclofenac sodium (DC-Na) and indomethacin (IDM)
aqueous g-cyclodextrin (gCD) solutions during from Sigma (St. Louis, MO), g-cyclodextrin (gCD)
storage11 and investigations of aqueous 1% (w/v) and 2-hydroxypropyl-g-cyclodextrin (HPgCD) MS
gCD solutions have shown that about 0.02% (w/w) 0.6 (MW 1576 Da) from Wacker Chemie (Munich,
of gCD in the solutions is in the form of Germany), disodium edetate dehydrate (EDTA)
aggregates.12 Comparable observations have and sodium chloride from Merck (Darmstadt,
been made in a-cyclodextrin (aCD) and b-cyclo- Germany), and benzalkonium chloride from
dextrin (bCD) solutions.13,14 The diameter of Sigma. Semi-permeable cellophane membranes
these cyclodextrin aggregates is approximately (SpectaPor1, molecular weight cut-off (MWCO)
100 nm. The cyclodextrin derivatives, such as 3500, 6 8000, 12 14,000) was purchased from
2-hydroxypropyl-b-cyclodextrin (HPbCD), are Spectrum Europe (Breda, Netherlands). All other
also thought to form aggregates in aqueous chemicals used were of analytical reagent grade
solutions.8 In aqueous solutions the relative purity. Milli-Q (Millipore, Billerica, MA) water was
amount of aggregated cyclodextrin increases with used for the preparation of all solutions.
increasing cyclodextrin concentration and per-
meation studies of hydrocortisone/HPbCD solu-
tions have indicated that in spite of the
Phase-Solubility Profiles
transparent nature of the solutions virtually all
increase in drug solubility observed at HPbCD The solubility of the drugs was determined by
concentrations above 10% (w/v) is due to hydro- previously described heating method.21 Hydro-
cortisone/HPbCD aggregate formation.8,15,16 cortisone, diclofenac sodium and indomethacin
However, aggregates of the natural aCD, bCD, were chemically stable (i.e., their degradation was
and gCD were studied in very dilute (about 1%, w/ less than 1%) during heating (1218C for 20 min) in
v) aqueous cyclodextrin solutions and the hydro- aqueous gCD and HPgCD solutions. However
cortisone/HPbCD aggregate formation was only amphotericin B was chemically unstable during
observed by permeation studies. In this present heating in an autoclave and thus amphotericin B
study three different techniques are used to verify solutions were saturated through heating in a
formation of the aggregates. Water-soluble poly- sonicator to 608C for 30 min. The suspension was
mers, metal ions and various carboxylic acids are heated to promote drug, and in some case
also known to influence the cyclodextrin solubi- cyclodextrin, saturation of the eye drop complexa-
lization of drugs, possibly through interactions tion medium. Excess amount of the drug to be
with cyclodextrin aggregates.17 19 tested was added to a solution containing 0 20%
Previously we have shown that addition of (w/v) cyclodextrin in eye drop vehicle consisting of
relatively small amounts of the water-soluble aqueous solution containing 0 20% (w/v) cyclo-
2-hydroxypropyl-g-cyclodextrin (HPgCD) to aqu- dextrin, benzalkonium chloride (0.02%, w/v),
eous gCD formulations increases the complexa- EDTA (0.1%, w/v) and sufficient sodium chloride
tion efficiency (CE) of gCD and reduces the to obtain isotonicity. The drug suspensions were
turbidity of gCD solutions. Mixing HPgCD with heated in an autoclave (1218C, 20 min), or in the
gCD resulted in synergistic effect with regard to case of amphotericin B to 608C for 30 min in a
drug solubilization in aqueous media.20 It was sonicator, and allowed to cool to room tempera-
hypothesized that the observed effect was some- ture. Then a small amount of solid drug was added
how related to formation of gCD and HPgCD to the suspensions to promote drug precipitation
nanoparticles. The purpose of this present study and the pH adjusted to 7.4 with concentrated
was to investigate further the solubilizing effects sodium hydroxide solution. The pH was periodi-
of gCD, HPgCD, and gCD/HPgCD mixtures and cally tested during the equilibration time and the
their formations of nanosize aggregates. samples were ready to be analyzed. The suspen-
sions were equilibrated at room temperature
(22 238C) for 7 days under constant agitation.
Preliminary experiments showed that 7 days are
MATERIALS AND METHODS
more than enough time to reach solubility
equilibrium. After equilibrium was attained, the
Materials
suspensions were filtered through 0.45 mm cellu-
Hydrocortisone (HC) was purchased from ICN lose membrane filter, the filtrates diluted with the
Biomedicals (Aurora, OH), amphotericin B (AmB), mobile phase and analyzed by HPLC. The
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 2, FEBUARY 2010 DOI 10.1002/jps
CYCLODEXTRINS AS SOLUBILIZERS: FORMATION OF COMPLEX AGGREGATES 721
apparent complexation constants for drug/cyclo- membrane. Before usage the membrane was
dextrin complexes (K1:1) were determined accord- soaked overnight in the receptor phase that
ing to the phase-solubility method of Higuchi and consisted of pH 7.4 phosphate buffer saline
Connors.22 The higher order complexation con- containing 5% (w/v) mixture of gCD/HPgCD (50/
stants were determined through nonlinear fitting 50 weight mixtures). To avoid possibility of
of the AP-type phase-solubility diagrams.22 24 The membrane fouling the receptor phase was auto-
complexation efficiency (CE) was determined from claved at 1218C for 20 min prior to use. Cyclodex-
the slope of linear phase-solubility diagrams (plots trin was added to the receptor phase to ensure
of the total drug solubility ([drug]t) versus total sufficient drug solubility to maintain sink condi-
cyclodextrin concentration ([CD]t) in moles per tion throughout the experiment. The receptor
liter):21 phase was sonicated under vacuum to remove
dissolved air before it was placed in the receptor
Slope ½drug=CD complexŠ
chamber. The donor phase (2 mL) consisted of
CE ź ź
1 Slope ½CDŠ
0 20% (w/v) cyclodextrin solution in eye drop
vehicle saturated with the drug, as previously
ź K1:1S0 (1)
described under methods in the phase-solubility
section. The study was conducted at room
where K1:1 is the stability constant of the drug/
temperature (22 238C) and under continuous
cyclodextrin 1:1 complex and S0 is the intrinsic
stirring for 6 h by a magnetic stirring bar rotating
solubility of the drug.
at 300 rpm. An aliquot of receptor medium
(150 mL) was withdrawn at 30, 60, 120, 180,
240, and 360 min and replaced immediately with
Quantitative Determinations an equal volume of fresh receptor phase. The drug
concentration in the receptor phase was deter-
The quantitative determinations of the individual
mined by HPLC. The steady state flux was
drugs were performed on a reversed-phase HPLC
calculated as the slope (dq/dt) of linear section
component system from Hewlett Packard Series
of the amount of drug in the receptor chamber
1100, consisting of a G132A binary pump with a
(q) versus time (t) profiles, and the apparent
G1379A solvent degasser, a G13658 multiple
permeability coefficient ( Papp) was calculated
wavelength detector, a G1313A auto sampler,
from the flux (J) according to Eq. (2):
and Phenomenex Luna 5 mm C18 reverse-phase
dq
column (150 mm 4.6 mm). The HPLC chromato-
J ź ź PappCd (2)
graphic conditions were shown in Table 1.
A dt
where A is the surface area of the mounted
membrane (1.77 cm2) and Cd is the initial con-
Permeation Studies centration of the drug in the donor phase.
The permeability of individual drugs was carried
out using Franz diffusion cell apparatus consisted
Determination of Activity Coefficients
of a donor and a receptor compartment (FDC
400 15FF, Vangard International, Neptune, NJ). The concentration dependences of osmolalities
The donor chamber and the receptor compartment were measured using Knauer K-7000 vapor
were separated by a semi-permeable cellophane pressure osmometer (Knauer, Germany). The
Table 1. HPLC Conditions
Drugs Mobile Phasea Flow Rate (mL/min) Wavelength (nm) Retention Time (min)
Hydrocortisone ACN:THF:water (33:1:66) 1.5 241 5.1
Diclofenac sodium ACN:1.0% acetic acid (60:40) 1.5 282 4.0
Indomethacin ACN:0.5% acetic acid (50:50) 1.5 240 6.9
Amphotericin B ACN:0.25 mM EDTA (37:63) 1.0 403 3.2
ACN, acetonitrile; THF, tetrahydrofuran; acetic acid, aqueous acetic acid solution; EDTA, aqueous disodium edetate dehydrate
solution.
a
Volume ratios.
DOI 10.1002/jps JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 2, FEBUARY 2010
722 JANSOOK, KURKOV, AND LOFTSSON
osmometer was calibrated with standard sodium these experiments. The process of preparation
chloride solutions of known osmolality. The was done under constant vacuum. Finally,
osmotic coefficients, w, were calculated from the samples were examined in a Model JEM-
osmolalities by equation: 2100 transmission electron microscope (JEOL,
Tokyo, Japan).
Osm ź n m (3)
where Osm is osmolality of a solute in osmol/kg, n
is the number of particles in which the solute
RESULTS AND DISCUSSION
molecules dissociate (i.e., n ź 1 when there is no
dissociation, like in our case), and m is the molal
The physicochemical properties of the drugs and
concentration of a solute.
cyclodextrins tested are shown in Table 2. All four
The activity coefficients, g, of solutes in the
drugs tested are somewhat lipophilic and water-
solutions studied were estimated using Bjerrum
insoluble with log Po/w ranging between 0.8 and
equation:
1.6 and solubility from about 1 mg/mL to 5 mg/mL
Zm at pH 7.4, whereas the cyclodextrins are much
more hydrophilic. The phase-solubility diagrams
ln g ź  1 þ ð 1Þ d ln m (4)
of the drugs were determined in the aqueous eye
0
drop medium, an aqueous solution containing
The concentration dependency of osmotic coeffi-
0 20% (w/v) cyclodextrin, 0.02% (w/v) benzalk-
cients is described by appropriate polynomial
onium chloride, 0.10% (w/v) EDTA and sufficient
function, that is:
sodium chloride to make the solution isotonic. All
the phase-solubility diagrams were of type A
 ź 1 þ B1m þ B2m2 þ B3m3 þ (5)
except the diagrams of hydrocortisone in gCD and
Then, Eq. (5) is introduced into Eq. (4) instead of w,
80/20 gCD/HPgCD solutions were of BS-type
which results in:
according to the Higuchi-Connor classification
system.22 The slopes of the initial linear sections
3 4
ln g ź 2B1m þ B2m2 þ B3m3 þ (6)
of the diagrams were determined and both the
2 3
apparent stability constants (K1:1) and the com-
As follows from Eqs. (5) and (6) the standard state
plexation efficiency (CE) determined from the
is represented by an ideal solution with m ! 0, for
slope (Tab. 3). In case of amphotericin B the values
which both w and g ! 1.
of K1:1 and K1:2 were determined by nonlinear
fitting to the AP profile.23 The slopes of the
diclofenac profiles were greater than unity indi-
Transmission Electron Microscopy (TEM) Analysis
cating that every drug/cyclodextrin inclusion
The morphology and size of the aggregates in complex contains one cyclodextrin molecule and
hydrocortisone saturated aqueous 10% (w/v) gCD, more than one diclofenac molecule or more likely,
HPgCD and gCD/HPgCD (80/20 and 20/80 weight that mixtures of inclusion and noninclusion
mixtures) solutions were analyzed using a trans- complexes are being formed.8,15,26 Mixtures of
mission electron microscope. One of the easiest gCD and HPgCD have synergistic solubilizing
ways of preparing the samples is by negative effects on hydrocortisone and diclofenac (Tab. 3).20
staining. This preparation method is useful for Both indomethacin and amphotericin B have very
visualizing suspensions of small particles. The low affinity for gCD and HPgCD as observed by
aggregates were visualized by TEM using the their very low CE value, indicating that only
uranyl staining method.25 However, the disad- about 2 7% of the cyclodextrin molecules in the
vantage of this method is possible aggregated solution are forming complexes with these drug
formation during sample preparation. Initially, molecules.21 The A-type phase-solubility dia-
formvar-coated grids were floated on a droplet of grams with slopes less than unity and the
the saturated preparation on parafilm, to permit observed stability constants indicate that in most
the adsorption of the nanoparticles onto the grid. cases one drug molecule forms a complex with one
After blotting the grid with a filter paper, the grid cyclodextrin molecule and that the molecular
was transferred onto a drop of the negative stain. weights (MWs) of the formed complexes are much
Following this, the grid was blotted with a filter less than 3500 Da. Even if a 2:1 diclofenac/
paper and air dried. Aqueous uranyl acetate HPgCD complex could be formed its MW would
solution (2%) was used as a negative stain in only be 2168 Da. The only exception is the 1:2
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 2, FEBUARY 2010 DOI 10.1002/jps
CYCLODEXTRINS AS SOLUBILIZERS: FORMATION OF COMPLEX AGGREGATES 723
amphotericin B/gCD and 1:2 amphotericin B/
HPgCD complexes with MW of 3518 and 4076 Da,
respectively. Thus, the complexes should be able
to permeate a semi-permeable cellophane mem-
brane with molecular weight cutoff (MWCO) of
3500 Da. It should also be remembered that in A-
type phase solubility diagrams, where aqueous
cyclodextrin solutions are saturated with poorly
soluble drug, the concentration of unbound drug is
constant and equal to the intrinsic solubility of the
drug in the aqueous complexation media. The
observed increase in drug solubility is due to
formation of water-soluble drug/cyclodextrin com-
plexes. Consequently hydrocortisone (MW 362.5)
permeation from aqueous hydrocortisone/HPbCD
solution through a cellophane membrane with
MWCO of 500 is constant and independent of the
total hydrocortisone solubility while the perme-
ability through comparable membranes with
MWCO of 6 8000 and 12 14,000 increase with
increasing solubility up to HPbCD concentration
of about 10% before leveling off.8,16 Figure 1 shows
the phase-solubility diagrams of the drugs in the
aqueous eye drop medium and their permeability
profiles from the eye drop medium through
MWCO 3500 semi-permeable cellophane mem-
brane. According to Fick s first law, the flux (J)
should be proportional to the concentration of
dissolved drug (Cd) in the complex medium (the
eye drop medium):
J ź PCd (7)
where P is the permeability coefficient. Linear
phase-solubility profile indicates that the concen-
tration of dissolved drug in the eye drop medium is
proportional to the cyclodextrin concentration
(Fig. 1). If the MW of the water-soluble drug/
cyclodextrin complex formed is less than 3500 Da
then a plot of J versus Cd should in theory
be linear. In other words if the phase-solubility
is linear and if the MW of the complex is
less than 3500 Da then a linear relationship
between the cyclodextrin concentration and
J through a MWCO 3500 membrane should be
observed. However, all the permeability profiles in
Figure 1 show negative deviation from linearity.
The profiles are linear up to cyclodextrin concen-
tration of about 5 10% (w/v). As the cyclodextrin
concentration increases the profiles start to level
off and eventually J becomes independent of the
amount of drug dissolved in the aqueous eye drop
medium for drugs that display type A phase-
solubility profiles, whereas J decreases for drugs
displaying BS-type profiles. For the water-soluble
DOI 10.1002/jps JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 2, FEBUARY 2010
OH
O
OH
OH
O
O
OH
O
O
OH
HO
O
a
HO
O
OH
OH
O
OH
OH
O
O
OH
OH
HO
O
OH
O
1576
>
500
<
10
O
200 (dec.)
OH
OH
O
HPgCD
O
O
O
O
O
O
O
HO
HO
HO
HO
OH
OH
OH
OH
HO
OH
O
O
HO
O
O
OH
HO
O
HO
O
OH
OH
O
O
OH
OH
12
OH
 
HO
249
O
gCD
O
1297.1
HO
200 (dec.)
O
OH
OH
HO
O
OH
O
O
O
OH
HO
O
HO
HO
HO
HO
COO
H
OH
H
O
OH
O
OH
2
HO
H N
OH
0.8
924.1
0.001
5.5; 10
>
170 (dec.)
Amphotericin B
O
H
O
OH
OH
OH
OH
O
O
OH
O
O
N
23,32 36
4.5
158
357.8
Cl
O
Cl
H
N
Cl
4.2
157
COOH
296.2
0.8 (at pH 7.4)
1.0 (at pH 7.4)
5.2 (at pH 7.4)
0.8 (at pH 7.2)
OH
O
H
HO
H
H

1.6
0.4
362.5
HO
Hydrocortisone
Diclofenac
Indomethacin
O
b
c
o/w
Hydroxypropylated gCD is a mixture of structurally related compounds. The chemical structure shown is only a representative structure.
Dec.
ź
decomposition upon heating.
The logarithm of the octanol/water partition coefficient.
a
b
c
a
(at RT)
0
Table 2.
Physicochemical Properties of the Sample Compounds
Physicochemical
Properties
Chemical structure
Molecular weight
Melting point (8C)
214 (dec.)
pK
log P
S
(mg/mL) in water
724 JANSOOK, KURKOV, AND LOFTSSON
Table 3. Apparent Stability Constant (K1:1, K1:2), and the Complexation Efficiency (CE) of Selected Drug/Cyclodextrin
Complexes in the Aqueous Eye Drop Medium at Room Temperature (22 238C)
Cyclodextrin Ratio Slopea Correlation Coefficient K1:1 (M 1) K1:2 (M 1) CE
Hydrocortisone
gCD  0.238 0.999 240  0.31
gCD/HPgCD 80/20 0.656 0.994 1460  1.90
gCD/HPgCD 20/80 0.617 0.996 1240  1.61
HPgCD  0.547 0.998 925  1.21
Diclofenac sodium
gCD  1.560 0.999  b   b
gCD/HPgCD 80/20 1.664 0.998  b   b
gCD/HPgCD 20/80 1.579 0.997  b   b
HPgCD  1.611 0.998  b   b
Indomethacin
gCD  0.028 0.944 25  0.03
gCD/HPgCD 80/20 0.024 0.966 21  0.02
gCD/HPgCD 20/80 0.031 0.972 28  0.03
HPgCD  0.035 0.944 31  0.04
Amphotericin B
gCD  0.063 0.982 1330 150 0.07
gCD/HPgCD 80/20 0.063 0.973 1180 170 0.07
gCD/HPgCD 20/80 0.048 0.960 1090 140 0.05
HPgCD  0.040 0.912 1060 120 0.04
The intrinsic solubility (S0) of hydrocortisone, diclofenac sodium, indomethacin and amphotericin B in the eye drop medium was
determined to be 0.47 mg/mL (1.3 mM), 6.18 mg/mL (19.4 mM), 0.41 mg/mL (1.2 mM), and 0.29 mg/mL (0.3 mM), respectively, at room
temperature.
a
For indomethacin this is the initial linear slope of the phase-solubility diagram. Hydrocortisone formed BS-type phase-solubility
diagrams with gCD and 80/20 gCD/HPgCD and thus the slope was calculated from the initial linear section of the profile.
b
Could not calculate. The slope of the phase-solubility diagram was greater than unity indicating that the complex contained more
than one drug molecule for every cyclodextrin molecule.
drug/cyclodextrin complexes (those that have type and gives profiles that level off when the
A profiles) virtually all increase in solubility at or cyclodextrin concentration approaches 5% (w/v).
above 10% (w/v) cyclodextrin concentration is The drug solubility decreases at cyclodextrin
due to formation of complex aggregates with MW concentration of about 10% (w/v) which is again
greater than about 3500 Da. From the phase- reflected in the permeability profiles. Indo-
solubility diagrams it can be estimated, based on methacin forms water-soluble complexes with
the hydrocortisone flux through the membrane, both pure gCD and the gCD/HPgCD mixture,
that in aqueous HPgCD and gCD/HPgCD solu- up to cyclodextrin concentrations 15 20% (w/v), as
tions saturated with hydrocortisone about 50% can be seen from the A-type phase-solubility
the hydrocortisone/cyclodextrin complexes are diagrams (Fig. 1C). Still the increase in J through
forming aggregates at 20% cyclodextrin concen- the MWCO 3500 membrane is insignificant and
tration, about 20% at 10% cyclodextrin concentra- relatively sharp break is observed in the linear J
tion and about 10% at 5% concentration. Thus, versus % cyclodextrin profile through the MWCO
complex aggregation appears to increase with 6 8000 and 12 14,000 membranes at 5% (w/v)
increasing cyclodextrin concentration. The flux of cyclodextrin concentration (Fig. 2). This indicates
hydrocortisone and indomethacin through semi- that although clear solutions are obtained after
permeable cellophane membranes with different filtration through 0.45 mm membrane filter, and
MWCO is shown in Figure 2. It can be seen that although A-type phase-solubility profiles are
the flux profiles of the hydrocortisone/gCD com- obtained, the water-soluble drug/cyclodextrin
plex is in agreement with the BS-type profiles complexes formed are unable to permeate mem-
(Fig. 1A) that level off at gCD concentration of branes with a MWCO of up to 14,000 Da. For
about 2.5% (w/v). The 80/20 gCD/HPgCD mixture hydrocortisone and indomethacin an aggregate
forms somewhat more water-soluble complexes with apparent MW of 14,000 Da will contain seven
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 2, FEBUARY 2010 DOI 10.1002/jps
CYCLODEXTRINS AS SOLUBILIZERS: FORMATION OF COMPLEX AGGREGATES 725
Figure 2. The effect of cyclodextrin concentration
on the flux of hydrocortisone (A) and indomethacin
(B) from the aqueous eye drop medium through
semipermeable cellophane membrane: MWCO
3500 (&); MWCO 6-8000 (*); and MWCO 12 14,000
(&). The cyclodextrin-containing aqueous eye drop
medium was saturated with the drug.
that contribution of drug/cyclodextrin dimers
to heptamers to the observed flux is negligible
although such aggregates are able to permeate the
MWCO 12 14,000 membrane.
The osmolalities of aqueous cyclodextrin solu-
tions were determined at room temperature
( 238C). The solutions contained 0 20% (w/v)
gCD, HPgCD or gCD/HPgCD mixtures (80/20 and
20/80 weight mixtures), or up to the saturation
region in the case of gCD and gCD/HPgCD (80/20).
Figure 1. Phase solubility profiles in the aqueous eye
drop media and the flux (J) profiles from the eye drop One experimental set consisted of pure aqueous
media through MWCO 3500 semipermeable cellophane
cyclodextrin solutions (Fig. 3A), while the other
membrane at room temperature. The cyclodextrin-
set contained 0.2 mg/mL hydrocortisone in iden-
containing eye drop media was saturated with the
tical cyclodextrin solutions (Fig. 3B). Some
drug: hydrocortisone (A); diclofenac sodium (B); indo-
literature data describing osmotic properties of
methacin (C); and amphotericin B (D). gCD (&); 80/20
the studied cyclodextrins are available. There is a
gCD/HPgCD (&); 20/80 gCD/HPgCD (*); and HPgCD
principal difference between our results and data
(*).
obtained by Miyajima et al.,27 which observed a
decrease in the value of osmotic and activity
to eight 1:1 drug/cyclodextrin complexes. Further- coefficients with increasing gCD concentration.
Our results are similar to those of Terdale
more, all the profiles in Figure 2 show a relative
et al.28 in case of aCD. However, an algorithm
sharp deviation from linearity at cyclodextrin
of osmotic coefficients calculation from vapor
concentration of about 5% (w/v) which indicates
pressure measurements was not described which
that complex monomers are in direct equilibrium
makes discussion of possible reasons of such
with aggregates containing not less than seven to
disagreement impossible. In addition, the osmotic
eight drug/cyclodextrin complexes and in similar
studies of HPgCD in a wide concentration range
fashion as when individual surfactant molecules
have been carried out by Zannou et al.29 Their
are in equilibrium will micelles. In other words,
DOI 10.1002/jps JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 2, FEBUARY 2010
726 JANSOOK, KURKOV, AND LOFTSSON
studied systems (Fig. 4A) which is in agreement
with published data for aggregation of aCD.28
The solutions can be arranged according to
increasing deviation from ideality: gCD < 80/20
gCD/HPgCD < 20/80 gCD/HPgCD < HPgCD. Such
deviations of activity coefficients from unity can
be due to either   solute solvent  interaction, that
is, hydration, or   solute solute  interaction,
that is, association or aggregation, or both. The
deviations from ideality are here mainly related to
the hydration process (namely, to hydrogen-bond
formation between oxygen atoms of the solute
and water, and hydrogen-bond formation in
the peripheral hydration) but the deviation is
also related, although to lesser degree, to solute
solute interactions.28,30 The hydration results in
increase in the activity coefficients but the
aggregation in a decrease. The observed increase
in the activity coefficients with increasing solute
concentration can be explained as follows. In
the aqueous cyclodextrin solutions a competition
between hydration and aggregation processes
is taking place where hydration prevails over
aggregation leading to increase in activity coeffi-
Figure 3. The experimental osmotic coefficients
cients above unity. Less positive deviation of
versus cyclodextrin molality at room temperature
activity coefficient is observed for pure gCD
( 238C). Cyclodextrin solutions without hydrocortisone
(Fig. 4A) and, in the context of competition, it
(A); cyclodextrin solutions containing 0.2 mg/mL of
means that the ratio between hydration and
hydrocortisone (B). gCD (^); gCD/HPgCD 80/20 (D);
aggregation terms is shifted in favor of the
gCD/HPgCD 20/80 ( ); HPgCD (&).
aggregation in comparison to the other systems.
The positive deviation increases with increasing
experimental osmolality values for 0 20% (w/v) gCD and HPgCD concentrations and the deviation
HPgCD lie between 0 and 200 mOsm/kg that is in is greater for HPgCD than for the parent gCD.
agreement with our observations. This is because HPgCD is a somewhat larger
The activity coefficients were estimated using molecule than gCD which allows for larger
Eq. (6) and data from Table 4. The concentration number of hydrogen-bonds (i.e., greater hydra-
dependences of activity coefficients for all studied tion) and consequently hydration/aggregation
solutions are shown in Figure 4. The solute ratio is shifted in favor of the hydration. Regard-
activity coefficients increase with concentration ing solutions containing cyclodextrins and hydro-
showing nonideal behavior of cyclodextrins in the cortisone the solutions can be arranged according
Table 4. Parameters of Eq. (5) for the Studied Aqueous Solutions at Room Temperature ( 238C)
Solution B1 B2 B3 R s 102 n
gCD 8 1 077 19 270 080 0.9913 1.89 7
80/20 gCD/HPgCD 10 2 104 27 383 121 0.9902 2.55 7
20/80 gCD/HPgCD 15 2 172 33 759 167 0.9948 2.49 7
HPgCD 18 1 214 18 995 095 0.9991 1.27 7
gCD þ HC 7 1 075 27  0.9858 1.00 6
80/20 gCD/HPgCD þ HC 12 4 587 225 6200 3300 0.9422 1.98 7
20/80 gCD/HPgCD þ HC 10 3 140 60 600 300 0.9363 4.43 6
HPgCD þ HC 11 2 111 39 370 200 0.9878 2.70 6
R, pair correlation coefficient; s, standard deviation; n, number of points.
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 2, FEBUARY 2010 DOI 10.1002/jps
CYCLODEXTRINS AS SOLUBILIZERS: FORMATION OF COMPLEX AGGREGATES 727
and 20/80 mixtures of gCD and HPgCD, saturated
with hydrocortisone are shown in Figure 5. The
sizes of hydrocortisone/gCD and hydrocortisone/
HPgCD complex aggregates are approximately
10 50 and 10 80 nm, respectively (Fig. 5A and B).
The aggregate diameter of the hydrocortisone/
(gCD/HPgCD) complex mixtures is much larger
than that of gCD and HPgCD or up to couple of
hundred nm in diameter, and appeared to be
formed by aggregation of smaller complex aggre-
gates (Fig. 5C and D). Previously we have shown
that addition of relatively small amounts of the
water-soluble HPgCD to aqueous gCD formula-
tions increases the CE of gCD and results in
synergistic effect with regard to drug solubiliza-
tion in aqueous media.20 It is possible that these
effects are related to formation of larger aggre-
gates. Other investigators have shown by TEM
imaging that the natural a-cyclodextrin (aCD),
b-cyclodextrin (bCD) and gCD, and their com-
plexes, form aggregates in pure aqueous solu-
tions.12 14,31 However, these were dilute
cyclodextrin solutions (1%, w/v) and the weight
Figure 4. The estimated activity coefficients of
fractions of the aggregates were only between
solute versus molality of the cyclodextrins studied in
0.0002% and 0.002% of the total cyclodextrin
aqueous solutions at room temperature ( 238C). Cyclo-
dissolved in the solution.13 Our permeation
dextrins without hydrocortisone (A); cyclodextrins with
studies indicate that the weight fraction of the
0.2 mg/mL of hydrocortisone (B). gCD ( ); gCD/HPgCD
aggregates gradually increases with increasing
80/20 (- - -); gCD/HPgCD 20/80 (- - -); HPgCD (. . .).
cyclodextrin concentration, especially when the
cyclodextrin concentration exceeds 5% (Figs. 1
to increasing deviation from ideality: 80/20 gCD/ and 2). We have made similar observations in
HPgCD < gCD < 20/80 gCD/HPgCD < HPgCD. aqueous HPbCD solutions.8,15,16
Figure 4B shows that the activity coefficients of all
tested drug containing cyclodextrin solutions
have in general smaller values than comparable CONCLUSIONS
solutions without the drug (Fig. 4A). This is due
to the fact that in addition to hydration and The natural aCD, bCD and gCD, their deriva-
aggregation the complexation process takes place tives, such as HPbCD and HPgCD, and complexes
in these drug containing solution systems. The self-assemble to form nanoscale aggregates in
complexation causes additional decrease in the aqueous solutions. At cyclodextrin concentration
value of the activity coefficients. This conclusion is of about 1% (w/v) or lower the relative mass
in agreement with complexation data from Table 3 contribution of these aggregates is less than
where the 80/20 gCD/HPgCD mixture has the 0.01% but it gradually increases with increasing
largest CE as well as the largest negative cyclodextrin concentration until at about 5 10%
deviation of activity coefficients (Fig. 4B). It would (w/v) when all increase in dissolved drug/cyclo-
be interesting to determine the driving forces of dextrin complexes is in the form of cyclodextrin
aggregation process via more detailed thermo- aggregates. In most cases these solutions do not
dynamic analysis that is by estimation of excess display static light scattering and appear to the
thermodynamic parameters of solutes (the Gibbs naked eye to be clear solutions and, thus, the
energy, enthalpy and entropy). This is a matter of diameter of the aggregates is below couple of
our future studies. hundred nanometers. Our studies indicate that
Transmission electron microscopic (TEM) most frequently the size is between 10 and 100 nm
micrographs of 10% (w/v) gCD and HPgCD which is in an agreement with previous studies by
solutions, as well as of aqueous 10% (w/v) 80/20 other investigators.13,14
DOI 10.1002/jps JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 2, FEBUARY 2010
728 JANSOOK, KURKOV, AND LOFTSSON
Figure 5. Transmission electron microscopic images of saturated solution of hydro-
cortisone in 10% cyclodextrin solutions: aggregated hydrocortisone/gCD complexes (A);
hydrocortisone/HPgCD complexes (B); hydrocortisone/(gCD/HPgCD) complexes with
gCD/HPgCD ratio of 80/20 (C); and hydrocortisone/(gCD/HPgCD) complexes with
gCD/HPgCD ratio of 20/80 (D).
ACKNOWLEDGMENTS 2. Dodziuk H, editor. 2006. Cyclodextrins and their
complexes. 1st edition. Weinheim: Wiley-VCH
Verlag.
Financial support from the Eimskip fund, Iceland,
3. Kajtár M, Vikmon M, Morlin E, Szejtli J. 1989.
is gratefully acknowledged. We gratefully
Aggregation of amphotericin B in the presence of
acknowledge the support of Scientific and Tech-
g-cyclodextrin. Biopolymers 28:1585 1596.
nology Research Equipment Centre, Chulalong-
4. Coleman AW, Nicolis I, Keller N, Dalbiez JP. 1992.
korn University, Bangkok, Thailand for TEM
Aggregation of cyclodextrins: An explanation of
analysis.
the abnormal solubility of b-cyclodextrin. J Incl
Phenom Macroc Chem 13:139 143.
5. Andronati SA, Shapiro YE, Yakubovskaya LN,
Gorbatyuk VY, Andronati KS, Krasnoschekaya
REFERENCES
SP. 1996. Inclusion compounds of psychotropic
agents and cyclodextrins. J Incl Phenom Macroc
Chem 24:175 186.
1. Loftsson T, Duchęne D. 2007. Cyclodextrins and
6. Gabelica V, Galic N, De Pauw E. 2002. On the
their pharmaceutical applications. Int J Pharm
specificity of cyclodextrin complexes detected by
329:1 11.
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 2, FEBUARY 2010 DOI 10.1002/jps
CYCLODEXTRINS AS SOLUBILIZERS: FORMATION OF COMPLEX AGGREGATES 729
electrospray mass spectrometry. J Am Soc Mass 23. Brewster ME, Loftsson T. 2007. Cyclodextrins as
Spectrom 13:946 953. pharmaceutical solubilizers. Adv Drug Deliv Rev
7. Correia I, Bezzenine N, Ronzani N, Platzer N, Beloeil 59:645 666.
J-C, Doan B-T. 2002. Study of inclusion complexes of 24. Saarinen-Savolainen P, Urtti A, Jarho P, Järvinen
acridine with b- and (2,6-di-O-methyl)-b-cyclodex- T. 1997. b-Cyclodextrin derivatives (2-HP-b-CD,
trin by use of solubility diagrams and NMR spectro- SBE4-b-CD) decrease the amphiphilicity and mem-
scopy. J Phys Org Chem 15:647 659. brane perturbing effects of pilocarpine prodrugs.
8. Loftsson T, Másson M, Brewster ME. 2004. Self- Eur J Pharm Sci 5:89 96.
association of cyclodextrins and cyclodextrin com- 25. Bugler J, Sommerdijk NAJM, Visser AJWG, van
plexes. J Pharm Sci 93:1091 1099. Hoek A, Nolte RJM, Engbersen JFJ, Reinhoudt
9. Bikádi Z, Kurdi R, Balogh S, Szemán J, Hazai E. DN. 1999. Interconnective host-guest complexation
2006. Aggregation of cyclodextrins as an important of beta-cyclodextrin-calix[4]arene couples. J Am
factor to determine their complexation behavior. Chem Soc 121:28 33.
Chem Biodiv 3:1266 1276. 26. Magnusdottir A, Másson M, Loftsson T. 2002. Self
10. Yan P, Tang J-N, Xiao J-X. 2007. Interaction association and cyclodextrin solubilization of
between b-cyclodextrin and mixed cationic-anionic NSAIDs. J Incl Phenom Macroc Chem 44:213 218.
surfactants (2): Aggregation behavior. J Dispers Sci 27. Miyajima K, Sawada M, Nakagaki M. 1983. Visc-
Technol 28:623 626. osity B-coefficients, apparent molar volumes, and
11. Szente L, Szejtli J, Kis GL. 1998. Spontaneous activity coefficients for a- and g-cyclodextrins in
opalescence of aqueous g-cyclodextrin solutions: aqueous solutions. Bull Chem Soc Jpn 56:3556
Complex formation or self-aggregation. J Pharm 3560.
Sci 87:778 781. 28. Terdale SS, Dagade DH, Patil KJ. 2006. Thermo-
12. Wu A, Shen X, He Y. 2006. Investigation of g- dynamic studies of molecular interactions in aqu-
cyclodextrin nanotube induced by N,N -diphenyl- eous alpha-cyclodextrin solutions: Application of
benzidine molecule. J Coll Interf Sci 297:525 533. McMillan-Mayer and Kirkwood-Buff theories.
13. Wu A, Shen X, He Y. 2006. Micrometer-sized rod- J Phys Chem B 106:18583 18593.
like structure formed by the secondary assembly of 29. Zannou EA, Streng WH, Stella VJ. 2001. Osmotic
cyclodextrin nanotube. J Coll Interf Sci 302:87 94. properties of sulfobutylether and hydroxypropyl
14. Bonini M, Rossi S, Karlsson G, Almgren M, Lo Nostro P, cyclodextrins. Pharm Res 18:1226 1231.
Baglioni P. 2006. Self-assembly of b-cyclodextrin in water. 30. Patil K, Pawar R, Dagade D. 2002. Studies of osmo-
Part 1. Cryo-TEM and dynamic and static light scattering. tic and activity coefficients in aqueous and CCl4
Langmuir 22: 1478 1484. solutions of 18-Crown-6 at 258C. J Phys Chem A
15. Loftsson T, MagnÅ›sdóttir A, Másson M, Sigurjóns- 106:9606 9611.
dóttir JF. 2002. Self-association and cyclodextrin 31. He W, Fu P, Shen XH, Gao HC. 2008. Cyclodextrin-
solubilization of drugs. J Pharm Sci 91:2307 2316. based aggregates and characterization by micro-
16. Loftsson T, Másson M, Sigurdsson HH. 2002. Cyclo- scopy. Micron 39:495 516.
dextrins and drug permeability through semi- 32. Moffat AC, Osselton MD, Widdop B, editors. 2004.
permeable cellophane membranes. Int J Pharm Clarke s Analysis of drugs and poisons. 3rd edition.
232:35 43. London: Pharmaceutical Press.
17. Loftsson T, Másson M. 2004. The effects of water- 33. Lavasanifar A, Samuel J, Kwon GS. 2002. The
soluble polymers on cyclodextrins and cyclodextrin effect of fatty acid substitution on the in vitro
solubilization of drugs. J Drug Del Sci Tech 14:35 43. release of amphotericin B from micelles composed
18. Redenti E, Szente L, Szejtli J. 2000. Drug/cyclodex- of poly(ethylene oxide)-block-poly(N-hexyl stea-
trin/hydroxy acid multicomponent systems. Proper- rate-L-aspartamide). J Control Rel 79:165 172.
ties and pharmaceutical applications. J Pharm Sci 34. Nokhodchi A, Javadzadeh Y, Siahi-Shadbad MR,
89:1 8. Barzegar-Jalali M. 2005. The effect of type and
19. Yamakawa T, Nishimura S. 2003. Liquid formula- concentration of vehicles on the dissolution rate
tion of a novel non-fluorinated topical quinolone, T- of a poorly soluble drug (indomethacin) from liqui-
3912, utilizing the synergic solubilizing effect of the solid compacts. J Pharm Pharmaceut Sci 8:18 25.
combined use of magnesium ions and hydroxypro- 35. Riess W, Stierlin H, Degen P, Faigle JW, Gerardin
pyl-b-cyclodextrin. J Control Rel 86:101 113. A, Moppert J, Sallmann A, Schmid K, Schweizer A,
20. Jansook P, Loftsson T. 2008. gCD/HPgCD: Syner- Sulc M, Theobald W, Wagner J. 1978. Pharmaco-
gistic solubilization. Int J Pharm 363:217 219. kinetics and metabolism of the anti-inflammatory
21. Loftsson T, Hreinsdóttir D, Másson M. 2005. agent Voltaren. Scand J Rheumato Suppl 22:17 29.
Evaluation of cyclodextrin solubilization of drugs. 36. Huang Y-B, Wu P-C, Ko H-M, Tsai Y-H. 1995.
Int J Pharm 302:18 28. Cardamom oil as a skin permeation enhancer
22. Higuchi T, Connors KA. 1965. Phase-solubility for indomethacin, piroxicam and diclofenac. Int
techniques. Adv Anal Chem Instrum 4:117 212. J Pharm 126:111 117.
DOI 10.1002/jps JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 2, FEBUARY 2010