Journal of Controlled Release 63 (2000) 305–317

www.elsevier.com / locate / jconrel

Investigation of the in vitro release of gentamicin from a

polyanhydride matrix
a, a a a a
Dennis Stephens *, Luk Li , Dan Robinson , Shen Chen , Hung-Chih Chang ,
Rong Ming Liu a , Youqin Tian a , Eric J. Ginsburg a , Xiaoyan Gao a , Timothy Stultz b
a
Advanced Drug Delivery, Hospital Products Division, Abbott Laboratories, Department 97 d, 100 Abbott Park Road, Abbott Park,
IL 60064 -3500, USA
b
University of Iowa, College of Pharmacy, Division of Pharmaceutics, Iowa City, IA 52240, USA

Received 26 June 1999; accepted 14 September 1999

Abstract

SeptacinE is a sustained release formulation consisting of gentamicin sulfate dispersed in a biodegradable polyanhydride
matrix. The polyanhydride matrix is a copolymer of erucic acid dimer (EAD) and sebacic acid in a 1:1 weight ratio. In vitro
drug release was performed in both water and pH 7.4 phosphate buffer. The drug release in water was faster than that in the
buffer, which was the opposite of what would be expected based upon a faster polymer hydrolysis rate in the buffer.
Theoretical treatment of the data using the Peppas model revealed that release in water was anomalous, while the release in
pH 7.4 phosphate buffer was diffusion-controlled. Profound bead morphology differences were observed between beads in
these two in vitro release media. Cracking was observed in beads in water and swelling with no apparent cracking was seen
in beads in buffer. Concurrent monitoring of drug and sebacic acid release indicated that drug release is not via surface
erosion. Osmotic effects were found to play little role in the in vitro drug release. There was no spectroscopic evidence of
amide formation between the drug and copolymer. Sulfate release was monitored along with drug release and the results
indicate that there is ion-exchange occurring during the pH 7.4 in vitro release. It was subsequently demonstrated that
gentamicin can form an insoluble salt with EAD. This salt formation explains the slower drug release in pH 7.4 phosphate
buffer.  2000 Elsevier Science B.V. All rights reserved.

Keywords: Polyanhydride; Gentamicin; In vitro release; Septacin

1. Introduction centration of the antibiotic at the site of infection by

Osteomyelitis is a deep bone infection that can
occur after hip or knee replacement surgery. Osteo-
myelitis is a particularly difficult infection to treat
because it is difficult to achieve a sufficient con-
*Corresponding author. Tel.: 11-847-938-2518; fax: 11-847-
938-3645.
E-mail address: stephend@hpd.abbott.com (D. Stephens)
systemic administration. This can be attributed to a
number of factors: the short half-life of the anti-
biotic; poor circulation to the infected area; and
systemic toxicity of the antibiotic, which prohibits
the use of a high systemic dose. Currently, osteo-
myelitis treatment requires large doses of antibiotics
that must be administered for long periods of time by
a combination of routes [1–3]. Some surgeons
currently blend antibiotics, such as aminoglycosides,

0168-3659 / 00 / $ – see front matter  2000 Elsevier Science B.V. All rights reserved.
PII: S0168-3659( 99 )00205-9
306 D. Stephens et al. / Journal of Controlled Release 63 (2000) 305 – 317
with poly(methyl methacrylate) (PMMA) to deliver
high local concentrations of drug [4]. A commercial
product consisting of PMMA beads loaded with
gentamicin has been approved for use in Europe [5].
PMMA suffers from the disadvantage that it is not
biodegradable and must be removed at a later date.
SeptacinE is a new product that is being de-
veloped for the treatment of osteomyelitis. It is a
controlled release implant that contains gentamicin
sulfate dispersed into a biodegradable polyanhydride
polymer matrix. Polyanhydrides [6,7] are a well-
studied class of bioerodible polymers [8–10]. The
drug loading is 20% (w / w) gentamicin sulfate. The
polyanhydride matrix is a copolymer of erucic acid
dimer (EAD) and sebacic acid in a 1:1 weight ratio.
Sebacic acid is the more hydrophilic monomer while
EAD is the more hydrophobic component. Drug
release can be tailored by varying the ratio of the two
monomers [6,7]. After incorporation of gentamicin in
the copolymer by melt-mixing, the drug polymer
blend is injection molded into a bead (12 mm33
mm) form that is suitable for use.
Septacin is designed to be implanted at the
surgical site when a hip or knee prosthesis is
replaced as a result of infection. The beads slowly
release gentamicin as the polymer degrades. This
provides a relatively high local concentration of drug
while minimizing systemic exposure. The efficacy of
polyanhydride-based delivery of gentamicin has been
shown in various osteomyelitis animal models
[11,12].
The purpose of this paper is to summarize the
investigation of the in vitro drug release of Septacin
in different dissolution media. Septacin, in essence,
is an extended release dosage form per the United
States Pharmacopeia definition [13]. In vitro drug
release, frequently run in pure water, is used for
these dosage forms as a quality control tool and to
demonstrate bioequivalence [14]. Septacin also
serves as a useful ‘model system’ for the release of a
hydrophilic drug from an extremely hydrophobic
polyanhydride matrix.
2. Materials
Septacin (Lots 15-944-DH, and 15-946-DH) and
Septacin Placebo (Lot 15-947-DH) were manufac-
tured by Abbott Laboratories (North Chicago, IL).
Sebacic acid (Lot 05418CQ) was obtained from
Aldrich Chemical (Milwaukee, WI). Gentamicin
sulfate (Lot 94224-KA-00) was obtained from Lek
Pharmaceuticals (Czechoslovakia). Erucic acid dimer
(EAD) (Lot 07904A500) was obtained from Unich-
ema (Chicago, IL).
3. Methods
3.1. In vitro dissolution
In vitro dissolution was performed by placing a
bead into a vial and adding 100 ml of dissolution
medium. The vial was stoppered and placed into a
reciprocal shaking water bath. The bath temperature
was maintained at 378C and the shaking speed was a
constant 100 rev. / min. The dissolution medium was
periodically changed by decanting the dissolution
contents and replenishing with a fresh 100 ml of
dissolution medium. The dissolution samples were
stored in a refrigerator prior to analysis. Samples
were typically collected after 1, 2, 3, 4, 5, 8, 11, 14,
18, 22, 30, and 37 days. All analyses were carried
out in triplicate by analyzing three beads in separate
vials. The percent cumulative drug released was
plotted against time in days and the error bars
represent 61 standard deviation.
3.2. Morphological evaluation
The morphology change of the Septacin beads
during the course of dissolution was investigated.
The beads were removed from the dissolution
medium and placed on filter paper. The samples
were photographed using a Sony DXC-97MD color
video camera with an 83 magnification. Remnants
were photographed after 2, 5, 11, 18, 26, and 34 days
of dissolution.
3.3. Analysis of gentamicin in in vitro dissolution
samples
Gentamicin sulfate concentrations of the in vitro
dissolution samples were determined using flow
injection analysis. Briefly, the sample is derivatized
with o-phthalaldehyde (FluoraldehydeE Pierce
 D. Stephens et al. / Journal of Controlled Release 63 (2000) 305 – 317
Chemical). The sample fluorescence is then mea-
sured using a flow injection system consisting of a
HPLC pump (Thermo Separation Products P4000), a
HPLC autosampler (Thermo Separation Products
AS3000), and a fluorescence detector (Thermo Sepa-
ration Products FL2000). Fluorescence was moni-
tored using an excitation wavelength of lex 5340 nm
and an emission wavelength of lem 5456 nm. The
sample response is quantitated against a linear
calibration curve that is generated using gentamicin
standard solutions.
3.4. Analysis of sebacic acid in in vitro dissolution
samples
Sebacic acid in the dissolution samples was
quantitated using a reversed phase HPLC with a
Waters Symmetry C-18 (5 mm) 2.1 mm3150 mm
column, and UV detection at 204 nm. The mobile
phase was 50% methanol / 50% 37 mM phosphate
buffer (pH53) maintained at a flow rate of 0.3
ml / min. The injection volume was 20 ml. Linearity
was established from 4 to 220 ppm sebacic acid.
Quantitation was performed using a single standard
whose concentration was 13 ppm. The HPLC system
consisted of a pump (Waters 410), autosampler
(Waters 717plus), and a UV detector (Applied Bio-
systems 785).
3.5. Sulfate determination in in vitro dissolution
samples
The release of sulfate from Septacin was moni-
tored by determining sulfate content of the dissolu-
tion medium. Sulfate was measured as sulfur using a
Perkin-Elmer 3000XL ICP fitted with a P-E cyclonic
spray chamber. The sulfur was determined at l5
180.669 nm using high purge and two point back-
ground correction (60.017 nm). A four-point cali-
bration was used with the sample matrix as zero
concentration and sulfur at 10, 30, and 50 mg / ml.
3.6. Gentamicin /EAD salt formation
The ionic interaction between gentamicin and
EAD was investigated as a function of pH. Portions
of EAD (100 mg) were placed in respective 125-ml
Erlenmeyer flasks. Gentamicin sulfate solutions (125
307
mg / ml) were prepared in 0.1 M phosphate buffers of
pH 6, 7, 8, and 9. Exactly 100 ml of the drug
solution were added to a flask containing EAD. The
mixture was immediately stirred at 500 rev. / min.
Triplicate samples were prepared at each pH. Con-
trols (drug solutions without EAD) were run under
the same conditions. After 48 h, a 0.5-ml aliquot was
filtered through a 0.45-mm Acrodisc  filter for
gentamicin analysis. The percentage of drug in
solution (relative to the respective control) was
calculated. Upon conclusion of the interaction study,
the pH of each flask was measured and then read-
justed to pH 5.0 using 5 N HCl. The amount of
gentamicin in solution was determined 24 and 48 h
after pH adjustment and the amount of gentamicin
recovered was calculated.
In a separate experiment, 2.26 g (|4.7 mmol) of
gentamicin free base [15] were dissolved in 100 ml
of methanol; 3.70 g (5.5 mmol) of EAD were
dissolved in 100 ml of methanol; 30 ml of EAD
solution were added to 15 ml gentamicin free base
solution resulting in a molar ratio of 2.3 mol EAD
per mol gentamicin, equivalent to the ratio of sulfate
to gentamicin in gentamicin sulfate. The solution
immediately became cloudy. The methanol was then
removed under vacuum. In vitro release of gen-
tamicin from the precipitate was determined in water
using the same in vitro dissolution procedure used
for Septacin.
3.7. Gentamicin /polysebacic acid amide formation
In order to search for any evidence of amide
formation between gentamicin and polyanhydride, an
attempt was first made to force amide formation
between polysebacic acid and gentamicin. This was
accomplished by mixing 159 mg of gentamicin
sulfate and 116 mg of polysebacic into 50 ml of
0.1 M pH 7.4 phosphate buffer and shaking at 100
rev. / min at 378C for 3 weeks. The mixture was
filtered, lyophilized, extracted with methanol, and the
methanol extract was concentrated to a solid. The
residue was subjected to spectroscopic analysis.
3.8. Spectroscopic analysis
Several spectroscopic methods were used to detect
formation of amide between gentamicin and polymer
308 D. Stephens et al. / Journal of Controlled Release 63 (2000) 305 – 317

in various samples. Infrared spectroscopy was per-

formed using a Nicolet Magna 750 FT-IR micro-
scope. Mass spectroscopy was performed using
several different techniques. Electrospray LC / MS
was performed using a Finnigan TSQ 7000 instru-
ment. Fast atom bombardment (FAB) mass spec-
troscopy was performed using a JEOL SX102 with
thioglycerol as the FAB matrix. Desorption chemical
ionization (DCI) was performed on a Finnigan SSQ
using ammonia as the chemical ionization reagent.
Atmospheric pressure chemical ionization (APCI)
mass spectrometry was performed using a Finnigan Fig. 1. In vitro release of Septacin in both water (♦) and
SSQ. Matrix assisted laser desorption ionization phosphate buffer at pH57.4 (j).
(MALDI) mass spectrometry was performed using a
Bruker Reflex unit with 2-(4-hydroxyphenylazo)- by a pure surface erosion mechanism, one would
benzoic acid (HABA) as the matrix. A liquid / liquid expect that release at pH 7.4 would be faster than
extraction was performed on Septacin remnant sam- that in more acidic, unbuffered water.
ples prior to mass spectral analysis. Remnants were
dissolved in 20 ml of methylene chloride and
extracted with an equal volume of 0.1 N H 2 SO 4 . 4.2. Theoretical treatment of release data
After extraction, there were also insoluble remnants
observed. Modeling of the controlled release of drugs from
Nuclear magnetic resonance (NMR) spectroscopy polymeric devices has been the subject of consider-
was also performed on various samples. Samples able research over the past 25 years. Most of the
were dissolved in either perdeuterated DMSO or models have been based on solutions of the Fickian
D 2 O. 13 C, 1 H, and heterocorrelation spectra were diffusion published in the classic book of Crank [18].
recorded using a Varian FT-NMR spectrometer. Higuchi derived a simple relationship that described
drug release from a matrix as a function of time [19]
(Eq. (1)).
4. Results and discussion
M
]t 5 kt 1 / 2 (1)
4.1. In vitro drug release in water and pH 7.4 M`
buffer
where Mt and M` are the respective masses of drug
The in vitro release of Septacin has been evaluated released at time equals t and `. It can be seen that
in both water (pH |4 to 5) and phosphate buffer
(pH57.4) and the release profiles are shown in Fig.
1. The observation that gentamicin was released
more quickly in the medium of lower pH (water) was
unexpected. Literature data suggest that the erosion
of polyanhydride copolymer (EAD:SA51:1 w / w) is
faster at higher pH [16,17]. In order to study the
copolymer erosion, degradation of the placebo beads
(copolymer without drug) was studied as a function
of pH. Sebacic acid release was monitored and the
results are summarized in Fig. 2. It is clear that the
release of sebacic acid is faster under more alkaline Fig. 2. Polymer erosion as a function of pH performed on the
conditions. Therefore, if drug release was controlled EAD:SA copolymer placebo.
 D. Stephens et al. / Journal of Controlled Release 63 (2000) 305 – 317
the relative release of the drug is linear with the
square root of time.
Peppas extended the Higuchi model to a more
generalized form:
M water. When the bead cracks, more surface area is
]t 5 kt n (2)
M`
where n is the diffusional exponent. Information
about the release mechanism can be gained by fitting
the drug-release data (for the first 60% dissolved)
and comparing the values of n to the semi-empirical
values for various geometries reported by Peppas
[20,21]. For a cylindrical geometry, values of n of
0.45 (or less) correspond to a purely Fickian diffu-
sion mechanism. Values of n greater than 0.89
indicate a relaxation controlled-release mechanism,
and n values between 0.45 and 0.89 indicate an
anomalous release mechanism.
The above equations are (at best) only approxi-
mations of a very difficult moving boundary-diffu-
sion problem. Assumptions made in the derivation of
these equations include pseudo-steady state approxi-
mations of Fick’s law, one-dimensional diffusion,
and no alteration of the polymer matrix. Despite the
approximations, it has been found that these relation-
ships can be applied to release data, indicating that
the diffusion-based models of Higuchi and Peppas
describe drug release from a polymeric system quite
well.
The data from the release curves of Fig. 1 were
fitted to the Peppas equation. The diffusional expo-
nent for the drug-release curve in water was 0.723,
indicating that the release mechanism was anomal-
ous. The release data in phosphate buffer, on the
other hand, yielded a diffusional exponent of 0.456,
indicating that Fickian diffusion plays an important
role. If drug release occurred purely by surface
erosion, as has been achieved for some poly-
anhydride / drug combinations [22], diffusion would
play little role. These results indicate that matrix
diffusion (not surface erosion) may play a significant
role in the release of gentamicin sulfate from Sep-
tacin in phosphate buffer at pH 7.4.
4.3. Changes in bead morphology
The morphology of the Septacin remnant was
309
monitored in both water and pH 7.4 phosphate
buffer. The results are summarized in Fig. 3. In
water, the bead was observed to crack over the
course of the drug release experiment. The cracking
may be attributable to the faster drug release in
exposed to water which facilitates faster drug release
through water-filled channels. This corresponds to
the anomalous release mechanism described by the
fit to Eq. (2). The beads in phosphate buffer do not
crack and appear to swell over the course of drug
release. This indicates that the matrix may serve as a
diffusion barrier and drug release is mainly achieved
by diffusion through the matrix. This result agrees
with the diffusion model as predicted by the fit to
Eq. (2).
The placebo bead in water (Fig. 3) shows no signs
of cracking indicating that the cracking of Septacin
in water is related to the presence of the gentamicin.
The placebo bead in phosphate buffer was found to
rapidly degrade. Based on this observation, it appears
that the gentamicin is slowing down the erosion of
the copolymer in the Septacin bead at pH 7.4. This is
based upon the fact that the Septacin bead remains
intact in pH 7.4 buffer while the placebo polymer
undergoes substantial erosion.
4.4. Osmotic effects
Osmotic forces have been shown in the literature
to substantially affect drug release for a variety of
different controlled release devices [23]. One could
theorize that the release of gentamicin would be
faster in pure water than in buffer because there is a
greater osmotic force driving water into the matrix as
the drug is dissolved locally in the polymer matrix.
This would create a relatively high local concen-
tration of drug within the matrix. This effect would
be less in the phosphate buffer because the dissolved
electrolytes in the buffer would result in a lower
osmotic pressure differential between the medium
and the matrix.
In order to investigate this effect, in vitro drug
release studies were carried out in 0.9% and 0.45%
NaCl. Sodium chloride solutions yield the same pH
as the water dissolution medium (pH55). The 0.1 M
pH 7.4 phosphate buffer was found to have an
osmolarity of 225 mOsm; 0.9% NaCl has a calcu-
310 D. Stephens et al. / Journal of Controlled Release 63 (2000) 305 – 317

Fig. 3. Morphological evaluation of bead remnants after 34 days in vitro dissolution: (a) Septacin in H 2 O, (b) Septacin in 0.1 M phosphate
buffer (pH57.4), (c) placebo polymer in H 2 O, (d) placebo polymer in 0.1 M phosphate buffer (pH57.4). Each square on the grid represents
a 333-mm area.

lated osmolarity of 303 mOsm, and 0.45% NaCl is

154 mOsm. The in vitro drug release of Septacin was
carried out in 0.9% and 0.45% NaCl and the release
curves (along with the water and pH 7.4 phosphate
buffer data) are plotted as Fig. 4.
It can be seen that the osmolarity of the dissolu-
tion medium has a relatively small effect on the drug
release rate. The most significant effect was realized
during the initial release. After |5 days, the release
curves for Septacin in 0.45% and 0.9% NaCl are
essentially parallel to the release curve in water. It
would appear that osmotic force plays a minor role, Fig. 4. Drug release of Septacin in pH 7.4 phosphate buffer (0.1
and it is not sufficient to account for the difference M), 0.9% saline, 0.45% saline, and water.
 D. Stephens et al. / Journal of Controlled Release 63 (2000) 305 – 317
seen in the drug release curves in phosphate buffer
and water. However, in comparison with the placebo
beads in water, the cracking of the Septacin beads in
water is likely attributed to the osmotic pressure
build-up by the dissolved gentamicin in the bead.
4.5. Release of sebacic acid
The polyanhydride utilized in Septacin is a co-
polymer of erucic acid dimer (EAD) and sebacic
acid. The sebacic acid is the more hydrophilic
monomer. It would be useful to compare the release
of this monomer with drug release in order to gain
some additional insight into the drug release mecha-
nism. In order to ensure that the sebacic acid release
profiles are obtained under sink conditions, the
solubility of sebacic acid in water and phosphate
buffer (pH57.4) at 378C was determined. The
solubility was determined to be 0.37 mg / ml in water
and 9.41 mg / ml in pH57.4 phosphate buffer. In all
cases the sebacic acid levels measured in the dissolu-
tion solutions were well below the measured solu-
bility limits. Gentamicin sulfate was also found to be
freely soluble (.2.8 mg / ml) in both water and
pH57.4 buffer.
The release of sebacic acid (in water) is plotted
along with the release of gentamicin in Fig. 5. It can
be clearly seen that the drug release and polymer
erosion (as shown by release of sebacic acid) curves
do not overlap. This indicates that the drug is not
released via a pure surface erosion mechanism and
that diffusion through the matrix is occurring. It
would appear that the drug is released faster than the
polymer erodes. This is due to the fact that gen-
Fig. 5. Release of drug and sebacic acid from Septacin in water.
311
tamicin is relatively hydrophilic. Shah et al. demon-
strated that the relationship between the drug release
rate and copolymer erosion rate is dependent on the
hydrophilic / hydrophobic nature of the loaded drug
[24]. They demonstrated that hydrophilic compounds
(such as mannitol) were released faster than the
polyanhydride erosion.
The release of sebacic acid is plotted along with
the release of gentamicin in pH 7.4 phosphate as
shown in Fig. 6. The release curves for the drug and
sebacic acid do not overlap, confirming that the drug
release in pH 7.4 buffer is also not via surface
erosion. The data are consistent with the results of
the theoretical treatment above. It is also noteworthy
that the sebacic acid release is apparently faster than
the drug release in pH 7.4 buffer.
The release profiles of sebacic acid from Septacin
in water and pH 7.4 buffer are comparable (Fig. 7).
This is very different from the results obtained with
the placebo beads, from which sebacic acid is
released approximately three times faster at pH 7.4
than at pH 5 (Fig. 2). This result indicates that the
release of sebacic acid is suppressed in the presence
of gentamicin. This could be due to a chemical
interaction between the drug and the copolymer and /
or its respective monomers.
4.6. Amide formation
One theory for the suppressed drug release of
Septacin in phosphate buffer is amide formation
Fig. 6. Release of drug and sebacic acid from Septacin in pH 7.4
phosphate buffer.
312 D. Stephens et al. / Journal of Controlled Release 63 (2000) 305 – 317
Fig. 7. Copolymer erosion of Septacin (as monitored by sebacic
acid release) in pH 7.4 phosphate buffer and water.
between the polyanhydride and the drug. Domb et al.
demonstrated that phenyl alkyl amines formed
amides with poly(sebacic acid) at pH 7.4 [25]. At pH
7.4 the amine groups on the gentamicin are suffi-
ciently deprotonated enabling nucleophilic attack at
Fig. 8. Desorption chemical ionization-mass spectrum (DCI-MS) of the methanol extract of mixing gentamicin sulfate and polysebacic acid
at pH 7.4. The spectrum is expanded 203 above m /z of 400.
the anhydride bond of the copolymer to form an
amide. The published pKa value for gentamicin is 7.9
[26]. Therefore, at pH 7.4 one can calculate (from
the Henderson–Hasselbach equation) that about 24%
of the available amines are deprotonated. Once the
amide is formed it may be insoluble, causing gen-
tamicin to be entrapped in the bead. Alternatively, if
a soluble amide is formed, the amine is blocked and
will no longer react with the o-phthalaldehyde. This
prevents its detection in the in vitro dissolution
samples.
The reaction between gentamicin and polysebacic
acid was studied as a model system. A desorption
chemical ionization mass spectral (DCI-MS) profile
(using ammonia as the CI reagent) of the extract is
shown in Fig. 8. It can be seen that the base ion in
the spectrum is at m /z of 220, which corresponds to
a [M1NH 4 ] 1 for sebacic acid. The structure of
gentamicin is given in Fig. 9. Gentamicin is a
 D. Stephens et al. / Journal of Controlled Release 63 (2000) 305 – 317
Fig. 9. The structure of gentamicin.
mixture of four components with molecular weights
of 477 (C1), 449 (C1a), and 463 (both C2 and C2a).
The three expected protonated molecular ions [M1
H] 1 are observed for the four components of gen-
tamicin. In addition, ions are observed at m /z of 662,
648, and 635 corresponding to [M1H] 1 of the
amides of gentamicin and sebacic acid.
The reaction mixture was also analyzed by NMR.
In the proton spectrum acquired in DMSO-d 6 , the
amide protons are observed from 7.5 to 8.5 ppm
indicating that some amide bonds are formed from
the reaction of one of the amine groups in gen-
tamicin with polysebacic acid. The amide protons
disappear when two drops of D 2 O are added to the
DMSO-d 6 solution. The amide carbonyl groups are
also observed in the carbon spectrum at 186.8 ppm.
However, the proton–carbon long range (2–3 bonds)
correlation spectrum did not reveal any correlation
between any proton in gentamicin and the carbonyl
groups. This result could be caused by the fact that a
small amount of amide bond is randomly formed at
the five possible sites in the gentamicin molecule so
that the concentration of each component is too low
to be detected.
Both NMR and mass spectroscopic analysis indi-
cate the gentamicin amide can be formed using
polysebacic acid in aqueous media. In Septacin,
gentamicin could be acylated with either EAD or
sebacic acid. Acylation by EAD should yield a
relatively insoluble amide. Remnants obtained after
in vitro drug release experiments were examined for
the presence of such amides.
313
After the liquid / liquid extraction of the remnant
samples, the methylene chloride layer, aqueous layer,
and the insoluble portion were all analyzed using a
variety of mass spectroscopic techniques (including
MALDI, FAB, and ESI) in both positive and nega-
tive ion modes. These analyses proved inconclusive,
as no amide was observed in any of the samples.
Infrared spectroscopy was also used to evaluate
the remnant samples. The infrared spectrum of the
insoluble portion of the remnant is presented in Fig.
10. The bands at 1642 and 1547 cm 21 could be
attributed to vibrations as a result of secondary
amide. However, the reference spectrum of gen-
tamicin sulfate (Fig. 11) also contains these bands.
Therefore, it could simply be that bands observed in
the remnant are from the residual drug. It was noted
that the prominent sulfate band in the gentamicin
sulfate reference spectrum at 1050 cm 21 was absent
in the insoluble remnant. This could possibly be the
result of ion-exchange of sulfate from the matrix to
the medium during the in vitro drug release. This
was further investigated by monitoring sulfate re-
lease in the medium.
4.7. Sulfate release
The release of sulfate as a function of dissolution
medium is shown in Fig. 12 along with the gen-
tamicin release data. It can be seen that the gen-
tamicin and sulfate release profiles are very similar
in water. In pH 7.4 phosphate buffer, at early time
314 D. Stephens et al. / Journal of Controlled Release 63 (2000) 305 – 317

Fig. 10. Infrared spectrum of insoluble remnant from phosphate in vitro dissolution.

points, sulfate release is significantly faster than
gentamicin suggesting that there is ion-exchange
occurring during the in vitro experiment [27].
These data are consistent with the hypothesis of
the formation of an insoluble salt of monomeric or
oligomeric carboxylate anion and protonated gen-
tamicin. As the polymer is hydrolyzed in pH 7.4
buffer, the resulting carboxylate anions can form an
ionic network with the multiply charged cationic
drug. The carboxylate exchanges for sulfate, which is
released into solution. In water the polymer hy-
drolysis is slower. This minimizes the amount of
carboxylate. In addition, the pH of the unbuffered
water is lower (approximately pH 5) so that the
carboxylate groups are protonated. This inhibits the
salt formation with gentamicin.
4.8. Gentamicin EAD salt formation
In order to further investigate the possibility of salt
formation between EAD and gentamicin, EAD was
added to a series of drug solutions at different pH.
The solutions were allowed to stir overnight in the
presence of the EAD and the gentamicin concen-
tration was measured. At pH greater than 5, the
equilibrium concentration of the gentamicin in the
aqueous phase decreases. This experiment essentially
is an indirect titration of the EAD and the results
indicate that the pKa of EAD is |6.5. It was also
demonstrated that when the pH was readjusted to
pH55, the gentamicin concentrations in the solu-
tions increased to the initial values.
In a separate experiment, the salt of gentamicin
 D. Stephens et al. / Journal of Controlled Release 63 (2000) 305 – 317
Fig. 11. Infrared spectrum of gentamicin sulfate.
free base and EAD was prepared. The salt was tested
for gentamicin release in water using the same
procedure used for the Septacin beads, except that on
the 21st day, a sufficient amount of H 2 SO 4 equiva-
Fig. 12. In vitro release of gentamicin and sulfate from Septacin
in both water and pH 7.4 phosphate buffer.
315
lent to 2.3 mol H 2 SO 4 per mol of gentamicin was
added to the dissolution solution. This ratio was
chosen because gentamicin sulfate contains 2.3 mol
of sulfate per mol of gentamicin. This lowered the
pH from |5.5 to 3.0. The results are summarized in
Fig. 13. It can be seen that before addition of H 2 SO 4 ,
the drug release profile is quite similar to that for
Septacin in pH 7.4 buffer. In addition, the lowering
of the pH after day 21 results in an increase in the
release rate. This is consistent with the carboxylate
ion becoming protonated, causing the dissociation of
the EAD and gentamicin, and release of soluble
gentamicin sulfate.
There is literature precedent for such salt forma-
tion. EAD salt formation in the absence of any drug
¨
has been investigated by Gopferich et al. [28]. They
showed DSC and IR evidence for the formation of
Na 1 EAD 2 salts during the dissolution of EAD/ SA
316 D. Stephens et al. / Journal of Controlled Release 63 (2000) 305 – 317
Fig. 13. In vitro release of gentamicin from the insoluble gen-
tamicin:EAD salt in water. On day 21, H 2 SO 4 was added to lower
the pH from 5 to 3.
polyanhydrides. In addition, salt formation between
gentamicin and oligolatic acid was reported by
Mauduit et al. [29]. Spectroscopic and release data
were accounted for by considering salt formation
between oligomer carboxylate chain ends and the
cationic form of the drug.
5. Conclusions
The results presented in this paper reveal the
complex nature of EAD/ SA dissolution and gen-
tamicin release from a polyanhydride matrix. Drug
release was faster in water than in pH 7.4 phosphate
buffer. This is contrary to what would be predicted
based upon the relative copolymer hydrolysis rates
and monomer solubilities in these two different
dissolution media. Fitting the in vitro drug release
data to the Peppas model indicated anomalous
release in water and diffusion controlled release in
pH 7.4 phosphate buffer. In addition, the beads
cracked in water, while cracking was not observed in
pH 7.4 buffer. Drug release was not governed by
surface erosion. In water, the release of drug was
faster than the copolymer erosion indicating that the
drug hydrophilicity and its diffusion through the
cracked matrix structure play a critical role.
Release in pH 7.4 buffer was found to be diffusion
controlled. The beads did not crack and were ob-
served to swell over the course of the experiment.
Osmotic effects of the dissolution medium were
found to be minimal. Amide formation between the
drug and the copolymer was also considered. Amides
were forced to form between gentamicin and poly-
sebacic acid p(SA). However, no amide formation
was observed between gentamicin and the EAD:SA
copolymer. Sulfate ion exchange was observed dur-
ing drug release at pH 7.4 indicating that a salt may
be formed between gentamicin and monomer (or
oligomers) in the degraded copolymer matrix. Addi-
tional data on the salt formation between gentamicin
and EAD further support the conclusion that the slow
release of gentamicin from Septacin in pH 7.4 buffer
is attributed to the formation of insoluble salt of
gentamicin and EAD monomer (or oligomers) in the
copolymer matrix.
These results indicate that the relationship between
the drug and the polymer can play a critical role in
the in vitro release characteristics. In addition,
cationic drugs have the potential to form hydro-
phobic salts with monomeric (or oligomeric) hy-
drolysis products of polyanhydrides. These salts can
radically affect the in vitro release profile.
References
[1] A.M. Ristuccia, B.A. Cunha, Antimicrobial Therapy, Raven
Press, New York, 1984.
[2] D. Schlossberg, Orthopaedic Infection, Springer-Verlag, New
York, 1988.
[3] R. Coombs, R.H. Fitzgerald, Infection in the Orthopaedic
Patient, Butterworths, Boston, 1989.
[4] R.P. Gruninger, D.T. Tsukayama, B. Wicklund, Suppressive
antibiotic therapy in chronic prosthetic joint infections, in:
R.B. Gustilo, R.P. Gruninger, D.T. Tsukayama (Eds.), Ortho-
paedic Infection, W.B. Saunders, Philadelphia, 1989.
[5] D. Seligson, S.L. Henry, Newest knowledge of treatment for
bone infection: antibiotic-impregnated beads, Clin. Orthop.
Relat. Res. 295 (1993) 2–118.
[6] J. Tamada, R. Langer, The development of polyanhydrides
for drug delivery applications, J. Biomater. Sci. Polymer
Edn. 3 (4) (1992) 315–353.
[7] P.A. Thomas, T. Padmaja, M.G. Kulkarni, Polyanhydride
blend microspheres: novel carriers for the controlled release
of macromolecular drugs, J. Control. Release 43 (2–3)
(1997) 273–281.
[8] M. Chasin, Biodegradable polymers for controlled drug
delivery, in: J.O. Hollinger (Ed.), Biomedical Applications of
Synthetic Biodegradable Polymers, CRC, Boca Raton, FL,
1995, pp. 1–15.
[9] T. Hayashi, Biodegradable polymers for biomedical uses,
Prog. Polym. Sci. 19 (4) (1994) 663–700.
[10] A.J. Domb, S. Amselem, Antibiotic delivery systems for the
treatment of chronic bone infections, in: A.J. Domb (Ed.),
Polymer Site-Specific Pharmacotherapy, Wiley, Chichester,
1994.
 D. Stephens et al. / Journal of Controlled Release 63 (2000) 305 – 317
[11] C.T. Laurencin, T. Gerhart, P. Witschger, R. Satcher, A.
Domb, A.E. Rosenberg, P. Hanff, L. Edsberg, W. Hayes, R.
Langer, Bioerodible polyanhydrides for antibiotic drug deliv-
ery: in vivo osteomyelitis treatment in a rat model system, J.
Orthop. Res. 11 (2) (1993) 256–262.
[12] C.L. Nelson, S.G. Hickmon, R.A. Skinner, Treatment of
experimental osteomyelitis by surgical debridement and the
implantation of bioerodable, polyanhydride-gentamicin
beads, J. Orthop. Res. 15 (1997) 249–255.
[13] United States Pharmacopeia, Vol. 22, Rand McNally, Taun-
ton MA, 1990, p. xliii.
[14] United States Pharmacopeia, Vol. 23, Rand McNally, Taun-
ton MA, 1995, p. lv.
[15] P. Singh, M. Pirio, D.K. Leung, Y.G. Tsay, Controlled
coupling of aminoglycoside antibiotics to proteins for use in
homogeneous enzyme immunoassays, Can. J. Chem. 62 (11)
(1984) 2471–2477.
[16] A.J. Domb, M. Maniar, Absorbable biopolymers derived
from dimer fatty acids, J. Polym. Sci.: Part A: Polym. Chem.
31 (5) (1993) 1275–1285.
[17] A.J. Domb, S. Amselem, J. Shah, M. Maniar, Poly-
anhydrides: synthesis and characterization, in: Advances in
Polymer Science, Vol. 107, Springer Verlag, Berlin, 1993, pp.
95–120.
[18] J. Crank, The Mathematics of Diffusion, Clarendon Press,
Oxford, 1975.
[19] T. Higuchi, Mechanism of sustained-action medication:
theoretical analysis of rate of release of solid drugs dispersed
in solid matrices, J. Pharm. Sci. 52 (6) (1963) 1145–1149.
[20] N. Peppas, P.L. Ritger, A simple equation for description of
solute release. I. Fickian and non-Fickian release from non-
swellable devices in the form of slabs, cylinders or disks, J.
Control. Release 5 (1) (1987) 23–36.
[21] N. Peppas, P.L. Ritger, A simple equation for description of
solute release. II. Fickian and anomalous release from
swellable devices, J. Control. Release 5 (1) (1987) 37–42.
[22] K.W. Leong, B.C. Brott, R. Langer, Bioerodible poly-
317
anhydrides as drug-carrier matrices. I: Characterization,
degradation, and release characteristics, J. Biomed. Mater.
Res. 19 (8) (1985) 941–955.
[23] X. Zhang, U.P. Wyss, D. Pichoria, M.F.A. Goosen, Bio-
degradable controlled antibiotic release devices for osteo-
myelitis: optimization of release properties, J. Pharm. Phar-
macol. 46 (9) (1994) 718–724.
[24] E.S. Park, M. Maniar, J.C. Shah, Influence of physiochemi-
cal properties of model compounds on their release from
biodegradable polyanhydride devices, J. Control. Release 48
(1) (1997) 67–78.
[25] A.J. Domb, L. Turovsky, R. Nudelman, Chemical interac-
tions between drugs containing reactive amines with hydro-
lyzable insoluble biopolymers in aqueous solutions, Pharm.
Res. 11 (6) (1994) 865–867.
[26] B.E. Rosenkrantz, J.R. Greco, J.G. Hoogerheide, E.M. Oden,
Gentamicin sulfate, in: K. Florey (Ed.), Analytical Profiles of
Drug Substances, Vol. 9, Academic Press, New York, 1980,
pp. 295–334.
[27] E. Ginsburg, T. Stultz, D. Stephens, D. Robinson, Y. Tian,
R.M. Liu, X. Gao, L. Li, J. Quick, H.C. Chang, An
investigation of the release properties of a cationic drug from
a hydrophobic polyanhydride matrix as a function of dissolu-
tion medium, in: T. Neenan, M. Marcolongo, R.F. Valentini
(Eds.), Biomedical Materials — Drug Delivery, Implants and
Tissue Engineering, Vol. 550, Proceedings from 1998 MRS
Fall Meeting, Boston MA, 1999, pp. 35–40.
[28] ¨
A. Gopferich, L. Schedl, R. Langer, The precipitation of
monomers during the erosion of a class of polyanhydrides,
Polymer 37 (17) (1996) 3861–3869.
[29] J. Mauduit, N. Bukh, M. Vert, Gentamycin / poly(lactic acid)
blends aimed at sustained release local antibiotic therapy
administered perioperatively. III. The case of gentamycin
sulfate in films prepared from high and low molecular weight
poly( DL-lactic acids), J. Control. Release 25 (1–2) (1993)
43–49.