International Journal of Pharmaceutics 164 (1998) 45–55
`
`Stabilisation of ionic drugs through complexation with
`non-ionic and ionic cyclodextrins
`
`Ma´r Ma´sson a, Thorsteinn Loftsson a,*, Sigrı´dur Jo´nsdo´ttir b, Hafru´n Fridriksdo´ttir a,
`Dorte Seir Petersen a
`a Department of Pharmacy, Uni6ersity of Iceland, IS-107 Reykja6ik, Iceland
`b Science Institute, Uni6ersity of Iceland, IS-107 Reykja6ik, Iceland
`
`Received 4 August 1997; accepted 17 November 1997
`
`Abstract
`
`The effects of negatively charged (i.e. carboxymethyl-i-cyclodextrin and sulfobutylether-i-cyclodextrin), positively
`charged (i.e. trimethylamoniumpropyl-i-cyclodextrin) and neutral cyclodextrins (i.e. hydroxypropyl-i-cyclodextrin,
`acetyl-i-cyclodextrin and randomly methylated i-cyclodextrin) on the chemical stability of various drugs were
`investigated. The degradation rate of each drug in aqueous cyclodextrin solutions was determined and the stability
`constant (Kc) of the drug–cyclodextrin complex and the degradation rate of the drug within the complex (kc) was
`obtained by non-linear fitting of the data. Compared to drug complexes with neutral cyclodextrins, the values of Kc
`were from 20 to 1600% larger when the drug and cyclodextrin molecules carried opposite charges, but from 50 to 80%
`smaller when the molecules carried the same type of charge. The values of kc were not affected by the charge of the
`cyclodextrin molecule. NMR studies of chlorambucil complexes indicated that the structure of the cyclodextrin
`complex was at least in some cases affected by the charge on the cyclodextrin molecules. © 1998 Elsevier Science B.V.
`All rights reserved.
`
`Keywords: Cyclodextrins; Ionic drugs; Complexation
`
`1. Introduction
`
`Cyclodextrins are cyclic oligosaccharides which
`are currently being investigated as pharmaceutical
`excipients, mainly as solubilizing and stabilising
`
`* Corresponding author. Fax: +354 5254071;
`thorstlo@hi.is
`
`e-mail:
`
`agents for lipophilic drugs in aqueous pharmaceu-
`tical formulations (Loftsson, 1995; Loftsson and
`Brewster, 1996). The cyclodextrin molecules have
`a hydrophilic outer surface and somewhat hydro-
`phobic central cavity. Many drugs are solubilized
`in cyclodextrin solutions through formation of
`drug–cyclodextrin inclusion complexes. Because
`of the nature of the cyclodextrin complex, a large
`
`0378-5173/98/$19.00 © 1998 Elsevier Science B.V. All rights reserved.
`PII S 0 3 7 8 - 5 1 7 3 ( 9 7 ) 0 0 3 8 7 - 6
`
`SENJU EXHIBIT 2024
`LUPIN v. SENJU
`IPR2015-01100
`
`Page 1 of 11
`
`

`
`46
`
`M. Ma´sson et al. /International Journal of Pharmaceutics 164 (1998) 45–55
`
`increase in the drug stability is frequently ob-
`served. However, in some cases the drug molecule
`interacts with the cyclodextrin hydroxyl groups in
`such way that the drug degradation is catalysed
`(Loftsson, 1995).
`The most common natural cyclodextrins are,
`h-, i- and k-cyclodextrins, which consist of 6, 7,
`and 8 glucose units, respectively. i-Cyclodextrin,
`and its derivatives, are the ones most commonly
`used for pharmaceutical applications since their
`central cavity has good affinity for many hydro-
`phobic structures of drug compounds (Fromming
`and Szejtli, 1994; Loftsson and Brewster, 1996).
`The parent i-cyclodextrin is not always ideal
`for drug formulations due to its moderate solubil-
`ity and reported toxicity after parental adminis-
`tration (Brewster et al., 1989). Therefore, various
`water/soluble i-cyclodextrin derivatives have
`been synthesised and used as pharmaceutical ex-
`cipients. The structure of many i-cyclodextrin
`complexes has been studied in detail by nuclear
`magnetic resonance (NMR) (Ueda and Nagai,
`1979; Loftsson et al., 1993). The available cy-
`clodextrin derivatives are not always suitable for
`such study since they consist of a mixture of a
`number of closely related derivatives and isomeric
`forms. It is often assumed that the nature of the
`cyclodextrin derivative complex is the same as
`that of the parent i-cyclodextrin complex,
`i.e.
`interaction between the drug and cyclodextrin is
`the binding in the hydrophobic cavity. However,
`it has been shown that cyclodextrin conforma-
`tions are modified to accommodate for methyl
`groups in methylated cyclodextrins, thus slightly
`changing the shape of the cyclodextrin cavity. It is
`therefore possible that the variation in the stabil-
`ity constant (Kc) and the degradation rate for the
`complexed drug (kc), can be explained by such
`changes in the shape of the cyclodextrin cavity.
`In the present work, we investigated the ionic
`interaction contribution to the complexation of
`the recently available,
`ionic cyclodextrins with
`ionic drug compounds. The values of both kc and
`Kc and degradation rate for the drug in buffer
`(ko), could be obtained from a series of degrada-
`tion studies. In this study, we used non-linear
`regression of the data rather than the previously
`reported linear regression (Loftsson, 1995) as this
`allowed better estimation of the error.
`
`2. Materials and methods
`
`2.1. Materials
`
`The cyclodextrins shown in Table 1 were used
`for
`this
`study. Carboxymethyl-i-cyclodextrin
`(CM-CD),
`trimethylamoniumpropyl-i-cyclodex-
`trin (TMA-CD), hydroxypropyl-i-cyclodextrin
`(HP-CD), acetyl-i-cyclodextrin (A-CD) and ran-
`domly methylated i-cyclodextrin (M-CD) were
`kindly donated by Wacker-Chemie (Germany),
`and
`sulfobutylether-i-cyclodextrin
`(SB-CD,
`MW2160) was kindly donated by CyDex (Kan-
`sas). The drug compounds were obtained from the
`following suppliers: acetyl salicyclate, salicylic
`acid, cephalotin and diazepam form Icelandic
`Pharmaceuticals (Iceland), indomethacin was pur-
`chased from Sigma Chemical Co. (USA), and
`chlorambucil was supplied by the courtesy of
`Wellcome Foundation Ltd. (UK). All other chem-
`icals were commercially available chemicals of
`reagent or analytical grade.
`
`2.2. Chromatography conditions for kinetic studies
`
`A stock solutions of cephalotin was made in
`water, acetylsalicylic acid in ethanol, and chlo-
`rambucil, diazepam, indomethacin and phenobar-
`bital in methanol. Between 10 and 7.5 vl of the
`drug stock solution were added to 1.5 ml of the
`cyclodextrin solutions, which were kept in a tem-
`perature controlled sample rack in an AS-4000
`(Merck-Hitachi) autosampler, and the changes in
`the drug concentration with time were monitored
`by HPLC. The HPLC system consisted of Con-
`stametric 3000 (Milton Roy) solvent delivery sys-
`tem with a SP8450 (Spectra-Physics) variable
`wavelength detector, using a 150-mm, 4.6-mm
`I.D., 5 vm bead, C18 reverse-phase column. The
`initial concentration of the drug in the reaction
`media was 3.2×10−5 M for cephalotin, 5.7×
`10−5 M for acetylsalicylic acid, 2.3×10−5 M for
`diazepam, 3.8×10−5 M for indomethacin, 5.7×
`10−5 M for phenobarbital and 3.4×10−5 M for
`chlorambucil. The mobile phases, detection wave-
`lengths and retention times for the different drugs
`were as follows: for cephalotin: acetonitrile/acetic
`acid/tetrahydrofuran/water
`(35:2:5:63 v/v), 260
`
`Page 2 of 11
`
`

`
`M. Ma´sson et al. /International Journal of Pharmaceutics 164 (1998) 45–55
`
`47
`
`Table 1
`Structure of the cyclodextrin derivatives
`
`Cyclodextrin derivative
`
`Supplier
`
`Substitution degree (DS)
`
`R
`
`CM-CD
`SB-CD
`HP-CD
`A-CD
`M-CD
`TMA-CD
`
`Wacker-Chemie
`CyDex
`Wacker-Chemie
`Wacker-Chemie
`Wacker-Chemie
`Wacker-Chemie
`
`0.5
`0.9
`0.6
`1.0
`0.6
`0.5
`
`–CH2COO−
`−
`–(CH2)4SO3
`–CH2CH(OH)CH3
`–COCH3
`–CH3
`+
`–(CH2)3N(CH3)3
`
`nm, 2.1 min; for chlorambucil: acetonitrile/acetic
`acid/water (55:1:44 v/v), 257 nm, 3.6 min; for
`diazepam: methanol/acetic acid/water (65:1:34 v/
`v), 228 nm, 3.8 min; for indomethacin: acetoni-
`trile/tetrahydrofuran/acetic
`acid/water
`(55:5:0.4:39.6 v/v), 256 nm, 3.4 min; and for phe-
`nobarbital: methanol/tetrahydrofuran/0.01 M
`phosphate (pH 7.7)/tetradecyltrimethyl ammo-
`nium bromide (51:5:44:0.02 v/v) 240 nm, 2.4 min.
`
`2.3. Data fitting
`
`The observed first-order rate constants in the
`aqueous cyclodextrin solutions (kobs) or ko for
`drug compounds other than diazepam was ob-
`tained from linear regression of the logarithm of
`the HPLC peak intensity plotted against time.
`The data was fitted using non-liner fitting of the
`Kaleidagraph program (Synergy Software, USA)
`
`which uses Levenberg-Marquardt algorithm for
`fitting of a user-defined equation. All data was
`fitted to a 1:1 complex model, according to the
`equation:
`ko+kc×Kc×[CD]
`1+KC×[CD]
`The values of kc and Kc were obtained from the
`best fit, but the ko was determined in aqueous
`buffer solutions containing no cyclodextrin.
`
`kobs=
`
`(1)
`
`2.4. NMR measurements
`
`in CH2Cl2
`A stock solution of chlorambucil
`was prepared. Sample (100 vl) of the stock solu-
`tion was added to a glass vial, the solvent evapo-
`rated under a stream of nitrogen and the residue
`dissolved in cyclodextrin containing D2O solution.
`The NMR spectra were recorded at 298 K in D2O
`
`Page 3 of 11
`
`

`
`48
`
`M. Ma´sson et al. /International Journal of Pharmaceutics 164 (1998) 45–55
`
`buffered solutions on a Bruker AC 250 spectrom-
`eter using standard software for water suppression.
`For calibration, the water signal was fixed at 4.80
`ppm. To diminish the hydrolysis of the drug, the
`NMR samples were prepared and equilibrated at
`25°C for 10 min just before the spectra were
`recorded. The chlorambucil concentration varied
`from 0 to 4 mM. For the spectra, scans from 50 to
`500 were necessary depending on the relative con-
`centration of the drug cyclodextrin concentration
`ratio.
`The observed change (Dobs) in the chemical shifts
`of the drug in cyclodextrin solutions was:
`Dobs=[D · CD]DCD/[D]t
`and the stability constant can be written as:
`[D · CD]
`[D][CD]
`
`Kc=
`
`=
`
`(2)
`
`([D · CD]/[D]t)
`(1−[D · CD]/[D]t)([CD]t/[D]t−[D · CD]/[D]t)([D]t)
`(3)
`
`By combining Eqs. (2) and (3), Eq. (4) was obtained
`where the negative solution had been discarded.
`([CD]tKc−Kc[D]t+1)
`2Kc[CD]t
`
`Dc
`Dobs=1+
`−
`([CD]tKc−Kc[D]t+1)2+4Kc[CD]t
`
`2Kc[CD]t
`
`Dc
`
`(4)
`
`In these equations, [D] is the concentration of
`free drug,
`[D · CD] concentration of complexed
`drug, [D]t is the total drug concentration, [CD] and
`[CD]t are the cyclodextrin and total cyclodextrin
`concentrations and Dc is the change in chemical
`shift of the drug when complexed with cyclodextrin.
`
`3. Results and discussion
`
`3.1. Relati6e affinity of the charged drugs for the
`cyclodextrin ca6ity
`
`The cyclodextrins had a marked effect on the
`degradation rate of chlorambucil,
`indomethacin
`
`and diazepam and thus the values of both kc and
`Kc could be calculated for these drugs. The cy-
`clodextrins had only minor effect on the degrada-
`tion of acetylsalicylic acid, phenobarbital and
`cephalotin. Previously, we have shown that acetyl-
`salicylate forms a complex with i-cyclodextrin at
`pH 1.0. (Loftsson et al., 1993). In our present study,
`we measured the degradation at higher pH (or pH
`7.0) in order to have the drug in fully ionised form.
`Ionic acetylsalicylate apparently has very little
`affinity for i-cyclodextrin cavity as no effect on the
`degradation rate could be observed for any of the
`cyclodextrins tested. NMR study of salicylic acid
`also confirmed the lack of complexation at pH 7.0.
`The degradation for negatively charged pheno-
`barbital was measured at 50°C in an aqueous 0.1
`M NaOH (pH 12.88), 75 mM cyclodextrin solu-
`tions. The rate of degradation was 25% slower in
`the aqueous SB-CD solution than in the pure buffer
`solution. The rate of degradation was 21% slower
`in aqueous HP-CD solutions, and 8% slower in the
`M-CD solutions, the effect was insignificant in the
`TMA-CD, and CM-CD solutions. Kc could not be
`estimated since the cyclodextrins had insignificant
`effect on the degradation rate below 10 mM cy-
`clodextrin concentration. The decreased rate of
`degradation did not reflect any significant ionic
`interaction. Also at such high cyclodextrin concen-
`trations, the effects observed could be due to
`secondary phenomena, such as changes in drug
`activity, rather than formation of a drug–cyclodex-
`trin complex.
`The experimental error (10%) in the determina-
`tion of kobs for cephalotin was to large to allow
`determination of the Kc values. Previously, it has
`been shown that cyclodextrin does form a complex
`with cephalotin at pH 6.5, but the difference
`between ko and kc was only about 10% (Loftsson
`and Johannesson, 1994).
`
`3.2. Non-linear fitting of the degradation data
`
`Fig. 1 shows the expected degradation pathways
`as previously reported (Connors et al., 1986) and
`the proposed structure of
`the
`complex for
`chlorambucil,
`indomethacin (Backensfeld et al.,
`1990) and diazepam. All the drugs are degraded
`via hydrolytic reaction in the aqueous solutions.
`
`Page 4 of 11
`
`

`
`M. Ma´sson et al. /International Journal of Pharmaceutics 164 (1998) 45–55
`
`49
`
`Fig. 1. Proposed structure of the drug–cyclodextrin complexes and the degradation pathways. (A) Chlorambucil, (B) indomethacin,
`and (C) diazepam. Diazepam is in equilibrium with its degradation product. The cyclodextrin is shown as a cylinder rather than a
`cone-shape, since it is frequently uncertain from which end the drug is entering the complex.
`
`Page 5 of 11
`
`

`
`50
`
`M. Ma´sson et al. /International Journal of Pharmaceutics 164 (1998) 45–55
`
`The overall observed degradation constant (kobs)
`depends on the degradation constant
`for the
`drug in solution (ko), and the degradation rate
`constant
`for drug within the
`complex (kc)
`and the stability constant of the drug–cyclodex-
`trin complex (Kc), as described by Eqs. (5) and
`(6):
`
`−
`
`d[D]t
`dt
`
`=kobs[D]t=ko[D]+kc[D · CD]
`
`Kc=
`
`[D · CD]
`[D][CD]
`
`=
`
`[D · CD]/[D]t
`(1−[D · CD]/[D]t)[CD]
`
`(5)
`
`(6)
`
`In these equations, [D] is the concentration of
`free drug,
`[D · CD] concentration of drug–cy-
`is the total drug con-
`clodextrin complex,
`[D]t
`centration,
`and
`[CD]
`and
`are
`the
`[CD]t
`cyclodextrin and total cyclodextrin concentra-
`tions [CD]:[CD]t. If [D]t[CD]t, then
`
`([D · CD]/[D]t)=
`
`Kc[CD]
`1+Kc[CD]
`
`(7)
`
`Finally, Eq. (8) can be obtained by combining
`Eqs. (5) and (7)
`
`3.3. Effects of cyclodextrins on the equilibrium
`between diazepam and its hydrolysis product
`
`The hydrolysis of diazepam is a reversible reac-
`tion. Whereas, chlorambucil and indomethacin
`are completely degraded in aqueous solution, an
`equilibrium will be established between diazepam
`and its degradation product. The degradation re-
`action is then described by the rate constants (k01
`and kc1,) for the hydrolysis reaction and the rate
`constants (k02 and kc2) for the dehydration reac-
`tion. Kc1 and Kc2 are the corresponding stability
`constants (Fig. 1.). Systems such as this have
`previously been described in some detail by
`Capellos and Bielski
`(1972). In this case,
`the
`concentration of the drug relative to the initial
`drug concentration is dependent on time accord-
`ing to the equation:
`kobs1+kobs2 e−(kobs1+kobs2)t
`[Diazepam]
`kobs1+kobs2
`[Diazepam]o
`where kobs1 and kobs2 are the observed rate con-
`stants for each cyclodextrin concentration and
`[Diazepam]o is the initial diazepam concentration.
`Fig. 3 shows that whereas indomethacin is de-
`
`(9)
`
`=
`
`kobs=ko1−
`
`
`
`1+Kc[CD]+kc Kc[CD]1+Kc[CD]
`
`Kc[CD]
`
`ko+kcKc[CD]
`1+Kc[CD]
`
`=
`
`(8)
`
`Both pH and the temperature of the reaction
`media were chosen such that
`the degradation
`rates of
`the drugs were between 65 and 2
`min−1, and the drug molecules were either fully
`positively charged or fully negatively charged.
`Fig. 2 shows how HP-CD affected the drug
`degradation. The graph shows kobs/ko versus
`[CD] with non-linear fitting, as this allows better
`visual
`comparison
`than
`the
`conventional
`Lineweaver-Burk method, where a linear regres-
`sion of ko/(ko−kobs) versus 1/[CD]
`is obtained
`(Loftsson, 1995). This method also puts equal
`weight on each data point, whereas the linear
`regression method will give increased weight to
`the lower concentration measurements, which
`can lead to larger errors, especially when Kc is
`small.
`
`Fig. 2. Degradation rate relative to the HP-CD concentration.
`() Chlorambucil, (“) indomethacin, ( ) rate of degradation
`of diazepam (k1), and (
`) rate of re-formation of diazepam
`(k2).
`
`Page 6 of 11
`
`

`
`M. Ma´sson et al. /International Journal of Pharmaceutics 164 (1998) 45–55
`
`51
`
`negative charge even at a low pH. Titration
`showed that CM-CD had lower pKa than the
`structurally similar carboxymethylcellulose,
`the
`respective values being 3.6 and 4.3 (Wade and
`Weller, 1994). This means that CM-CD is fully
`ionised above pH 5.
`Table 2 shows the Kc and kc value for the drug
`compounds with the various cyclodextrins. Nega-
`tively charged chlorambucil had less affinity for
`the two negatively charged cyclodextrins than for
`the neutral cyclodextrin, but the largest complexa-
`tion was obtained with the positively charged
`TMA-CD. The stability constants of
`the in-
`domethacin–cyclodextrin complexes were also
`smaller in the case of the negatively charged cy-
`clodextrins than in the case of the neutral cy-
`clodextrins.
`The opposite trend was observed with the posi-
`tively charged diazepam, but, because of small
`value of Kc1, it was difficult to interpret the data.
`The degradation product formed more easily a
`complex with cyclodextrins (larger Kc2 values) and
`in this case the Kc2 observed with SB-CD, which
`was the only cyclodextrin negatively charged at
`pH 2, was 5–17 times larger than with the neutral
`cyclodextrins and 9–3 times smaller with TMA-
`CD. Comparable observations have been made
`for SB-CD by other investigators (Okimoto et al.,
`1996)
`
`3.5. Complexation of indomethacin with
`TMA-CD
`
`Complexation of indomethacin with TMA-CD
`was somewhat different from the other drug–cy-
`clodextrin complexes studied. In the beginning,
`when the cyclodextrin concentration was
`in-
`creased the degradation rate decreased, but then it
`increased again upon further increase in the cy-
`clodextrin concentration (Fig. 4.). These changes
`could not be due to ionic strength changes since
`the ionic strength was equally high in other solu-
`tions containing ionic cyclodextrins in which the
`data was consistent with 1:1 complex formation.
`It has been shown that in the same solution there
`can co-exist more than one form of cyclodextrin
`inclusion complexes (Aki et al., 1996; Crouzy et
`
`Fig. 3. Degradation profile for indomethacin (A) and di-
`azepam (B) in CM-CD solutions. Diazepam: () pure buffer
`solution, (“) 5 mM CM-CD, ( ) 10 mM CM-CD, (
`) 20
`mM CM-CD, () 35 mM CMCD, and () 75 mM CM-CD.
`Indomethacin: () pure buffer solution, (“) 3.3 mM CM-CD,
`( ) 8 mM CM-CD, (
`) 20 mM CM-CD, () 35 mM
`CM-CD, and () 75 mM CM-CD solution. The buffer values
`were average of four measurements.
`
`graded exponentially, the diazepam degradation
`progressed towards an equilibrium. In this case,
`cyclodextrin influenced the equilibrium between
`diazepam and the hydrolysis product to favour
`the latter compound, and therefore the initial rate
`of degradation increased with increasing cyclodex-
`trin concentration. Fig. 2 shows that the rigid
`diazepam structure complexed poorly, whereas
`the more flexible degradation product formed cy-
`clodextrin complex more easily.
`
`3.4. Cyclodextrin charge and the stability
`constant for the complex
`
`Table 1 shows the structures of the cyclodextrin
`derivatives. TMA-CD is positively charged and
`SB-CD is a strong acid and carries permanent
`
`Page 7 of 11
`
`

`
`52
`
`M. Ma´sson et al. /International Journal of Pharmaceutics 164 (1998) 45–55
`
`Table 2
`Stability (Kc) and degradation constants (kc) for the drugs in complex with cyclodextrin derivatives
`
`Cyclodextrin
`
`Charge
`
`Kc (×10−3, [M−1])
`
`kc [min−1]
`
`kc/k
`
`R 2
`
`Chlorambucil (40°C, pH 7.35)
`CM-CD
`SB-CD
`HP-CD
`A-CD
`M-CD
`TMA-CD
`
`Indomethacin (40°C, pH 9.8)
`CM-CD
`SB-CD
`HP-CD
`M-CD
`TMA-CD
`Diazepam (30.5°C, pH 2.0) KC1 and kc1
`CM-CD
`SB-CD
`HP-CD
`M-CD
`TMA-CD
`Diazepam (30.5°C, pH 2.0) KC2 and kc2
`CM-CD
`SB-CD
`HP-CD
`M-CD
`TMA-CD
`
`* ko for the drug at the indicated conditions.
`
`−
`−
`−
`0
`0
`0
`+
`
`−
`−
`−
`0
`0
`+
`
`+
`0
`−
`0
`0
`+
`
`+
`0
`−
`0
`0
`+
`
`0.7290.03
`1.4090.05
`3.4090.72
`2.6290.04
`3.5590.53
`4.2590.36
`
`0.0790.01
`0.2690.02
`0.6990.06
`0.7790.09
`Did not form 1:1 complex
`
`65.391.4*
`3.290.6
`1.090.4
`3.691.0
`2.190.1
`3.390.6
`1.290.4
`8.5390.36*
`1.0990.51
`−0.0990.23
`0.8490.13
`1.4290.20
`
`0.0190.00
`0.0790.06
`No effect
`0.1490.05
`No effect
`
`0.3690.07
`2.0590.26
`0.1290.01
`0.18+0.01
`0.0490.01
`
`2.590.1*
`25.096.9
`1.790.2
`
`3.090.1
`
`6.090.4*
`2.290.2
`0.090.1
`0.590.1
`0.590.1
`1.890.2
`
`0.049
`0.015
`0.055
`0.032
`0.051
`0.018
`
`0.12
`−0.01
`0.09
`0.16
`
`10.0
`0.7
`
`1.2
`
`0.37
`0.00
`0.08
`0.08
`0.30
`
`0.99
`1.00
`0.79
`1.00
`0.88
`0.95
`
`0.97
`0.98
`0.97
`0.96
`
`0.98
`0.73
`
`0.86
`
`0.91
`0.92
`1.00
`0.99
`0.99
`
`al., 1996). In such cases, the observed values of Kc
`and kc are the weighted averages of the different
`forms as long as all the complexes are of 1:1
`stoichiometric ratio. If an 1:2 complex (where the
`drug molecule is complexed with two cyclodextrin
`molecules) is also formed it will dominate at higher
`cyclodextrin concentrations. If the charge–charge
`interactions are weak in the normal
`inclusion
`complex, such interaction could be increased by
`interaction with another cyclodextrin molecule (i.e.
`formation of 1:2 complex). This could explain the
`kobs pattern seen for the indomethacin TMA-CD
`solutions. At higher concentration, a 1:2 complex
`could be formed with higher kc values than the 1:1
`complex predominant at lower concentrations.
`
`3.6. The kc 6alues for drugs in a complex
`
`cyclo-
`relationship between the
`No direct
`dextrin charge and the kc could be observed.
`However, it was notable that kc for SB-CD was
`always lower than for the other cyclodextrins
`and in the case of indomethacin and diazepam
`(i.e.
`for kc2) not
`significantly different
`from
`zero. SB-CD is more highly substituted than the
`other CD molecules, with the exception of A-
`CD, with the charged groups extended far away
`from the
`cyclodextrin cavity. The
`environ-
`ment around the
`cyclodextrin cavity could,
`therefore, offer better protection of
`the drug
`molecule.
`
`Page 8 of 11
`
`

`
`M. Ma´sson et al. /International Journal of Pharmaceutics 164 (1998) 45–55
`
`53
`
`3.7. NMR of chlorambucil complexes
`
`The calculated NMR shifts for fully complexed
`molecules are shown in Tables 3 and 4. When
`chlorambucil formed a complex with the i-cy-
`clodextrin the protons which are inside the cavity
`are shifted downfield, mainly due to the an-
`isotropical effect from the aromatic moiety inside
`the cavity. Our measurements showed that the
`shift was largest for H%-5, and slightly larger shift
`for H%-6, than H%-4. This places the benzene ring
`near the narrow cavity (H%-6 end). The small
`shifts of chlorambucil H-1, H-2 and H-3 protons
`showed that these were outside the cavity. The
`aromatic protons are shifted downfield as they
`will in hydrophobic environment. The H-6 and
`H-7 protons, forming one singlet, will be inside
`the cavity and, thus, they were shifted upfield. The
`hydrophobic effect shifts the protons downfield
`but the data obtained in a CDCl3/F3CCOOH
`media indicates that stronger hydrogen bonding
`of the nitrogen ion pair will have the opposite
`effect. The data therefore indicates that chloram-
`bucil forms stronger hydrogen bonding inside the
`cavity, with a co-complexed water molecule or
`cyclodextrin –OH group, than outside the cavity.
`The lowest energy structure of the complex would
`therefore be as shown in Fig. 1, with the benzene
`ring closer to the narrow end of the cavity.
`
`Fig. 4. Degradation profile for indomethacin in TMA–CD
`complex. The data could be fitted for 1:1, 1:2 complex system.
`
`The cyclodextrin derivatives are mixtures iso-
`mers and, thus, all the NMR peaks obtained were
`broad. The NMR shifts of the derivatives could
`therefore not be studied, but the chlorambucil
`peaks were sharp. The cavity of the cyclodextrin
`derivatives should have similar structure as that of
`the parent i-cyclodextrin and similar shifts would
`therefore be
`expected for
`the
`chlorambucil
`molecule. The shifts for CM-CD, SB-CD, M-CD
`and HP-CD are similar to those obtained for
`i-cyclodextrin. That is a −0.163 to −0.204 ppm
`shift of H-5 and approximately 30% larger shift of
`H-4 (i.e. −0.222 to −0.281) and a positive shift
`of H-6,7 which confirmed that the position of
`chlorambucil within all these complexes was the
`same.
`TMA-CD has a positive charge and therefore
`the most favourable position of the negatively
`charged acid group of chlorambucil
`is not far
`away from the positively charged cyclodextrin.
`The most favourable position of the drug should
`therefore be expected to be different from the
`other cyclodextrin complexes. NMR confirmed
`this prediction as indicated by the H-4 and H-5
`shifts. This was also observed for A-CD, with
`lower shifts, a effect which could not be predicted.
`Hydrogen bonding could play some role in this
`complex formation.
`Degradation of the other drugs in A-CD com-
`plexes was not studied as A-CD was prone to
`hydrolysis at higher pH and elevated tempera-
`tures, forming acetic acid which affected the pH.
`
`4. Conclusion
`
`The ionic cyclodextrins form stronger com-
`plexes with counter-ionic drugs but weaker com-
`plexes with drugs carrying the same type of
`charge, compared to comparable complexes with
`non-ionic cyclodextrins. The charge–charge forces
`were usually not additive to the other forces in-
`volved in the complex formation. If the forces had
`been additive, one would expect at least a 10-fold
`increase in the value of the stability constant, i.e.
`Kc, which was not the case. The NMR studies
`indicate that this lower than expected increase in
`the stability constant could be due to changes in
`
`Page 9 of 11
`
`

`
`54
`
`M. Ma´sson et al. /International Journal of Pharmaceutics 164 (1998) 45–55
`
`Table 3
`Shifts of the i-cyclodextrin protons when complexed with chlorambucil relative to buffer and the relative shifts of the chlorambucil
`protons
`
`i-Cyclodextrin
`
`Chemical shift
`
`Relative shift
`
`Proton
`
`H%-1
`H%-2
`H%-3
`H%-4
`H%-5
`H%-6
`Chlorambucil
`
`Proton
`
`H-1
`H-2
`H-3
`H-4
`H-5
`H-6
`H-7
`
`Buffer
`
`5.096
`3.993
`3.907
`3.890
`3.674
`3.609
`
`Free
`
`0.000
`0.000
`0.000
`0.000
`0.000
`0.000
`
`In complex
`
`−0.028
`−0.018
`−0.068
`−0.002
`−0.164
`−0.087
`
`Chemical shift
`
`Relative shift
`
`Buffer
`
`2.208
`1.828
`2.556
`7.215
`6.885
`3.757
`3.757
`
`Buffer
`
`Methanol
`
`0.000
`0.000
`0.000
`0.000
`0.000
`0.000
`0.000
`
`0.105
`0.085
`0.026
`−0.113
`−0.161
`−0.041
`−0.041
`
`CD2Cl2
`
`0.163
`0.094
`0.034
`−0.130
`−0.220
`−0.099
`−0.099
`
`CD2Cl2/F3CCOOH
`
`0.279
`0.191
`0.223
`0.254
`0.584
`0.275
`−0.186
`
`* The H%-6 proton of i-cyclodextrin was observed as broad singlet. The relative shift of the H%-5 was calculated from the two peaks
`which were readily observable close to the H%-6 peak. The H-6 and 7 protons of chlorambucil formed a singlet in buffer and
`cyclodextrin solutions but a multiplet or two triplets in other solutions.
`
`the location of the drug molecule within the cy-
`clodextrin cavity. That is, when the cyclodextrin
`and drug molecules carry opposite charges, the
`drug molecule has to arrange itself within the
`cavity to allow for the ionic interactions, but at
`the same time the forces between the drug
`molecule and the cyclodextrin molecule within the
`cavity will be reduced. For example, in some cases
`the hydrophobic moiety of a drug molecule,
`which under normal conditions would be located
`well inside the cavity, will partly be located out-
`
`side the cavity to allow for ionic interactions
`between the drug and the cyclodextrin molecules.
`In the case of indomethacin and TMA-CD, the
`ionic interaction and the drug–cyclodextrin inter-
`actions within the cavity appeared to be more or
`less competing with each other not allowing the
`two forces to operate simultaneously in a 1:1
`complex.
`No relationship could be observed between the
`charge on the cyclodextrin molecule and the rate
`of drug degradation within the cyclodextrin cavity
`
`Page 10 of 11
`
`

`
`M. Ma´sson et al. /International Journal of Pharmaceutics 164 (1998) 45–55
`
`55
`
`Table 4
`Shifts (in ppm) of chlorambucil in cyclodextrin complexes relative to buffer
`
`Proton
`
`Cyclodextrin
`
`B-CD
`
`CM-CD
`
`SB-CD
`
`HP-CD
`
`M-CD
`
`A-CD
`
`TMA-CD
`
`0.02290.006
`0.03290.004
`0.00690.003
`0.02490.001
`H-1
`−0.03590.004 −0.02590.005
`−0.02490.001
`ND*
`H-2
`H-3 −0.01390.003 −0.01690.001 −0.02790.004 −0.00890.002
`H-4 −0.22290.028 −0.28190.014 −0.24790.009 −0.23090.017
`−0.17790.021 −0.20190.010 −0.16390.008
`−0.17290.013
`H-5
`H-6,7
`0.05790.003
`0.03990.003
`0.06790.007
`0.07790.004
`
`0.08090.004
`0.00190.003
`−0.05190.005
`ND
`−0.03390.003
`0.05490.003
`−0.26690.013 −0.12190.004
`−0.20490.010
`−0.11990.004
`0.06990.004
`0.03790.002
`
`0.03790.000
`0.03790.000
`0.01790.000
`−0.20790.002
`−0.20590.002
`0.04890.002
`
`* Could not be determined due to overlapping cyclodextrin peaks.
`
`(i.e. the value of kc). However, the value of kc was
`generally smaller in the SB-CD solutions which
`indicates that the drug molecules were better pro-
`tected against hydrolysis within the cavity of SB-
`CD than within the
`cavity of
`the other
`i-cyclodextrin derivatives tested.
`
`Acknowledgements
`
`The work was supported by grant from the
`University of Iceland Research Fund.
`
`References
`
`Aki, H., Haraguchi, T., Niiya, T., Yamamoto, M., 1996.
`Multiple modes for inclusion complexation between phe-
`nobarbital and 2-hydroxypropyl-i-cyclodextrin in aqueous
`solution. Yakugaku Zasshi 116, 803–812.
`Backensfeld, T., Muller, B.W., Wiese, M., Seydel, J.K., 1990.
`Effect of cyclodextrin derivatives on indomethacin stability
`in aqueous solution. Pharm. Res. 7, 484–490.
`Brewster, M.E., Simpkins, J.W., Hora, M.S., Stern, W.C.,
`Bodor, N., 1989. The potential use of cyclodextrins in
`parenteral formulations. J. Parent. Sci. Technol. 43, 231–
`240.
`Capellos, C., Bielski, B.H.J., 1972. Kinetic Systems. Wiley-In-
`terscience, New York.
`
`Connors, K.A., Amidon, B.L., Stella, V.J., 1986. Chemical
`Stability of Pharmaccuticals. Wiley, New York.
`Crouzy, S., Fauvelle, F., Debouzy, J.-C., Goschl, M.,
`Chapron, Y., 1996. Investigation of the h-cyclodextrin-
`myo-inositol phosphate inclusion complex by NMR spec-
`troscopy and molecular modeling. Carbohydr. Res. 287,
`21–35.
`Fromming, K.H., Szejtli, J., 1994. Cyclodextrins in Pharmacy,
`vol. 5. Kluwer, Dordrecht.
`Loftsson, T., 1995. Effects of cyclodextrins on the chemical
`stability of drugs in aqueous solutions. Drug Stabil. 1,
`22–33.
`Loftsson, T., Brewster, M.E., 1996. Pharmaceutical applica-
`tions of cyclodextrins. I. Drug solubilization and stabiliza-
`tion. J. Pharm. Sci. 85, 1017–1025.
`Loftsson, T., Jo´hannesson, H.R., 1994. The Influence of cy-
`clodextrins on the stability of cephalotin and aztreonam in
`aqueous solutions. Pharmazie 49, 292–293.
`Loftsson, T., O´ lafsdo´ttir, B.J., Fridriksdo´ttir, H., Jo´nsdo´ttir,
`S., 1993. Cyclodextrin complexation of NSAIDS: physico-
`chemical characteristics. Eur. J. Pharm. Sci. 1, 95–101.
`Okimoto, K., Rajewski, R.A., Uekama, K., Jona, J.A., Stella,
`V.J., 1996. The interaction of charged and uncharged
`drugs with neutral (HP-i-CD) and anionically charged
`(SBE7-i-CD) i-cyclodextrins. Pharm. Res. 13, 256–264.
`Ueda, H., Nagai, T., 1979. Nuclear magnetic resonance
`(NMR) spectroscopy of inclusion compounds of tolbu-
`tamide and chlorpropamide with i-cylodextrin in aqueous
`solution. Chem. Pharm. Bull 28, 1415–1421.
`Wade, A., Weller, P.J., 1994. Handbook of Pharmaceutical
`Exicipients. American Pharmaceutical Association, Wash-
`ington.
`
`Page 11 of 11