Mechanical deformation of polymer matrix controlled release devices modulates drug release Elazer R. Edelman: Anthony Fiorino,' Alan Grodzinsky,' and Robert Langer' Cardiovascular Division, Department of Internal Medicine, Brigham and Women2 Hospital and Harvnrd Medical Schools, Boston, Massachusctts 02115; Harvard-M.I.J. Division of Health Sciences and Technology, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139; 'Dcpartnient of Chemical Engineerinx, Massachusetts Institute vf Technology, Cambridge, Massachusetts 02139; 'Department of Electrical Engineering and Computer Science, Massachusetts institute of Technology, Cambridge, Massachusctts 02139 When magnetic fields were applied t o polymer matrices of ethylene-vinyl acetate copolymer embedded w i t h d r u g a n d a small magnet, drug release was increased u p to 30-fold a b o v e baseline levels. I t has been hypothe5ized that the effect of magnetic stimulation on the release of d r u g s from these matrices is the transduction of the applied magnetic field into a mechanical deformation of the matrix through motion of the magnet within the matrix. This current study provides support for this hypothesis by demonstrating
that repeated pulsatile mechanical deformation of matrices can enhance the release of macromolecules from ethylenevinyl acetate copolymer matrices. Furthermore, similar modulated release kinetics were obtained with mechanically compressed and magnetically stimulated matrices. We also established that modulation was dependent on the ratio of compression area to matrix volume and that modulation was maximized when this ratio was optimized. 0 1492 John Wiley & Sons, Inc.
INTRODUCTION
Controlled release of a variety of compounds has been achieved with polymer-based drug delivery systems.' External modulation of this release has been demonstrated in a number of systems. When magnetic fields were applied to polymer matrices of ethylene-vinyl acetate (EVAc) embedded with drug and a small magnet, drug release was increased up to 30-fold above baseline This phenomenon was hypothesized to involve a mechanical interaction between the applied field and the polymer-drug matrix. Indeed, magnified video images of the surface of an EVAc matrix embedded with a magnet showed alternating compression and decompression of the matrix material adjacent to the magnet during exposure of the matrix to an oscillating magnetic field.3 A preliminary theoretical construct involving mechanical deformation of EVAc-drug matrices followed mass transfer between two solute reservoirs *To whom correspondence should be addressed. Journal of Biomedical Materials Kesearch, Vol. 26, 1619-1631 (1992) CCC 0021-9304/92/121619-13 0 1992 John Wiley & Sons, Inc.
EDELMAN ET AI,.
1620
connected by a long thin tube. This geometry was used to model the drug release from the network of caves and connecting channels that compose the polymer-drug matrices. Solute transfer rate between the reservoirs was obtained by solving the diffusion equation at points far from the ends of the tube, using the velocity profile established by an oscillating piston compressing the end of one tube. The increase in mass transfer above diffusion was found to be a function of three dimensionless groups: an oscillating Reynolds number, an amplitude factor relating the displacement of the piston to the radius of the tube, and the Schmidt n ~ m b e r . ~ In this study we investigated the potential of precisely controlled mechanical compression of EVAc-drug matrices to modulate drug release in practice. This precise control permitted us to investigate whether the same parameters that were important in affecting magnetic controlled modulation played a similar role in modulation obtained with direct mechanical compression and to investigate how the matrix itself might influence the extent of release modulation.
MATERIALS A N D METHODS
Materials Bovine serum albumin (BSA, Sigma Chemical Co., St. Louis, MO, M, 68,800) was sieved to particle sizes between 180 and 250 pm using standard USA testing sieves (ASTME specifications, Dual Manufacturing Co., Chicago, IL). Ethylene-vinyl acetate copolymer (EVAc, 40% vinyl acetate), manufactured under the product name ELVAX-40P, was washed in distilled water and 95% alcohol to remove impurities, and then dissolved in methylene chloride to form a 10% weight by volume s ~ l u t i o n . ' - Magnetized ~.~ stainless steel spheres (80% iron, 17% chromium) were obtained from Ultraspherics Inc. (Ann Arbor, MI). Matrix fabrication Matrices were fabricated in an identical fashion to those used in previous controlled release BSA was added to the EVAc solution to make a 33% weight by volume suspension, vortexed, let stand for 15 s to allow air bubbles to settle out and then poured into glass molds that had been precooled on dry ice. When magnets were embedded within the matrices they were dropped into the hardening EVAc: BSA solution 15 s after pouring. This left the magnets to reside in the center of the matrix. Once hardened, the matrices were removed from their molds, placed in a -20°C freezer for 2 days and then placed under house vacuum (600 mTorr) for another 2 days. The resulting matrix was a homogeneous dispersion of BSA within EVAc. Different sized and shaped glass molds were used to cast cylindrical matrices of varying heights and diameters. When placed in solution predictable and repro-
MECHANICAL DEFORMATIOK O F EVAc:BSA MATIIICES
1621
ducible release kinetics were ob~erved.'-~,~ To assure that we were operating within the linear range of release all matrices were soaked in 0.1M phosphate buffer for 72 h. This allowed for the surface BSA to elute off of the matrix surface and eliminated the burst phase of release. Controlled compression Our model reconstructed the magnetic system by examining the interface between the magnetic spheres and the polymer matrix. In the magnetic system compression occurred within the center of the matrix. As this was inaccessible we applied controlled compression at the surface of the matrices. We hoped to determine whether mechanical compression might produce the same kinetics of release, verifying the validity of the model, and thereafter permit us to examine the importance of various parameters effecting modulatable release. A Dynastat dynamic mechanical spectrometer (IMASS, Inc., Hingham, MA) was used to impose sinusoidal compressive displacements of EVAc matrices over a range of frequencies. The instrument is described in detail elsewhere.(' A special compression chamber was constructed from plexiglass in the shape of a cup with a raised pedestal in the center [Fig. l(a)]. The chamber was 3 cm in diameter and 2.5 cm high and the pedestal, 1.25 cm high and 1.5 cm in diameter. The matrices were placed on the pedestal and release of BSA was allowed to reach to equilibrium after the chamber was filled with 8 mL of 0.1M phosphate buffer at pH 7.4. Compression was applied to the matrix face in one of four different configurations. In the first three, one of different radii spherical magnets (0.7, 1.4, and 3.0 mm) used in the magnetic system was placed between the plunger and the matrix [Fig. l(b)]. In the fourth case the spheres were removed and the samples were then compressed by a solid plexiglass plunger fixed to the upper jaw of the Dynastat. The plunger face was 2 cm in diameter and covered the entire matrix, thus, in the latter case the samples were subjected to uniaxial unconfined compression while completely immersed in buffer. These four configurations modeled the situation wherein magnets of increasing size or of greater ability to affect modulation were used. Mechanical matrix stimulation with both uniaxial compression and point deformation involved a static offset displacement, to keep the matrix in position, and a superimposed adjustable dynamic displacement for potential modulation of BSA release. Magnetic field application Magnetic fields were generated from two different devices.23In the first, permanent magnets were secured to a plate rotating beneath the matrices. Magnetic fields ranged from 0 to 60 Hz at 1100 G peak to peak. In the second an electromagnetic field was used (plate demagnetizer, O.S. Walker Co., Inc., Worcester, M A ) capable of delivering a 60-Hz magnetic field at amplitudes
EDELMAN ET AL.
1622
MATRIX
I
RESTRAINING
CEIl.
-I
-..:
MAGNET
Figure 1. EVAc:BSA matrices were subjected to controlled mechanical deformation. The matrices were placed on a raised pedestal in the center of a plexiglass chamber that could be filled with fluid. RSA was released into the fluid at steady state. As the matrix was deformed by the plunger release rates increased in a manner dictated by the deforming force and frequency. In separate experiments the plunger was either in (a) direct contact with the matrix face or (b) a magnetic bead of variable size was placed in between. The mechanical force in the former experiments was referred to as uniasinl compression and in the latter a s point deformation.
varying from 0 to 900 G (peak to peak). The amplitude of the magnetic field was set to a value capable of displacing the embedded magnet 50 Fm. Release kinetics
Release was followed continuously or at selected intervals. For continuous monitoring of BSA release from the matrices a flow-through spectrophotometer was employed in the open loop c~nfiguration.~ lnlet and outlet ports of the compression chamber allowed for a continuous pumped flow of phosphate buffer (pH 7.4) over the matrix. The outlet solution was pumped through the spectrophotometer, where its absorption was recorded at 280 nm.
MECHANICAL DEFORMATION OF EVAc: BSA MATRICES
1623
The recorded data was then digitized and stored on a computer (SUN Microsystems, Mountain View, CA) for later analysis. System dynamics were defined in the following fashion. Phosphate buffer was pumped through the spectrophotometer cell and at various times BSA solutions of known concentrations were injected into the cell, and the resulting changes in absorbance recorded. Mixing of the BSA solution with the phosphate buffer in the cell and tubing established an exponential rise and decay in absorbance. In subsequent analyses of the matrix response to mechanical or magnetic stimulation the system rise and decay values were used to remove the secondary effects introduced by the flow-through system. This was accomplished by digitally cross-correlating the stimulated response to the system response. In order to follow the BSA release periodically, the solution bathing the matrix was sampled after time intervals of 5 or 10 min. This was accomplished by draining the buffer from the cell and storing it in a glass scintillation vial; the cell was then refilled with 8 mL of fresh buffer. The absorbance at 280 nm was measured to determine the amount of BSA present in each sample. The extent of release enhancement, “modulation,” was defined as the ratio of mass transfer of drug out of the matrix during stimulation to the basal rate of release. This was calculated by dividing the absorbance attained during the application of a dynamic deformation by the baseline absorbance measured in the absence of any oscillating deformation.
RESULTS
System dynamics Solutions of known concentrations of BSA injected through the flowthrough circuit into the compression chamber appeared as a rise in absorbance after 2 min; this delay reflected the transit time through the tubing connecting the chamber and the spectrophotometer. The plateau and baseline BSA concentrations determined from the absorbance recorded in both the flow-through and periodic configurations identically matched the known injected BSA concentration ( p < 0.0001,R = 0.999), indicating no loss of BSA in the sample chamber or circuit tubing. The system response was determined for the controlled injection of five BSA concentrations from 0.1 to 0.5 mg/mL. Matrix deformation and dynamics of release enhancement To determine the validity of our model we examined the BSA release profile from EVAc matrices subjected to mechanical or magnetic stimulation (Fig. 2). A mechanical compressive displacement of 50 pm was applied to the matrix, and compared to the results with application of a 60-Hz magnetic field at an amplitude capable of deforming an identical matrix by displacing the internal magnetic bead by 50 pm. Both curves demonstrate negligible
EDELMAN ET AI..
360
720
1080
1440
Time (seconds) Figure 2. BSA release kinetics from a n EVAc matrices subjected to mechanical deformation were virtually identical to the kinetics that followed magnetic field stimulation of an EVAc: BSA matrix embedded with a magnetic sphere. A 14-min matrix stimulation was applied at 0 time in each case. A 50-pn deformation of the matrix was applied at 60 Hz and compared to a 60-Hz oscillating magnet field capable of displacing the matrix material adjacent to the embedded magnet 50 k m as well. Though different degrees of modulation were observed the amount of BSA released was normalized to allow for direct visual comparison of the kinetics of the two different means of matrix stimulation.
baseline BSA release which increased almost immediately with mechanical deformation or magnetic field exposure. After a stable plateau was obtained and the stimulus was removed, release returned to baseline almost immediately. The rise and decay time constants associated with initiation and termination of matrix stimulation were obtained for both mechanical compression and magnetic field stimulation of EVAc matrices (Table I) and matched identically ( p < 0.0001). Mechanical deformation When the EVAc-BSA matrices were exposed to dynamic uniaxial compression, modulation increased linearly with frequency (5 to 60 Hz, Fig. 3; slope
MECHANICAL DEFORMATION OF EVAc: BSA MATRICES
1625
TABLE I Time Constants of EVAc: BSA Matrix Response Kinetics to Mechanical and Magnetic Stimulation Derived from Curves Depicted i n Figure 2 Time Constant (ms)
Mechanical Magnetic
Kise
Decay
87.41 87.34
71.94 73.53
12
10
4
2' 3
I
10
I
2 0
I
I
I
4 0
50
6 0
1
30
FREQUENCY (Hz) Figure 3. The extent of modulation of BSA release from EVAc matrices is plotted against the frequency of mechanical matrix deformation (closed circles) and magnetic field oscillation (open circles). Identical EVAc : BSA matrices were used. In the magnetic experiment a spherical magnet was embedded within the EVAc:BSA matrix. The matrix was then exposed to a 60-Hz oscillating magnetic field capable of displacing the matrix material around the magnet 50 pm. Mechanical studies were conducted with the magnet placed between the deforming plunger and the matrix; the plunger was set to provide a 50-pm displacement at 60 Hz. A linear increase in modulation was observed for both types of stimulation and within the same range for both magnetic (slope 0.12 ? 0.02, SEE = 0.80, R = 0.99, p = 0.09), and mechanical (slope 0.13 2 0.01, SEE = 0.12, R = 0.99, p = 0.0002) forms of matrix release manipulation.
70
EDELMAN El' AI,.
1626
0.13 % 0.01, SEE = 0.12, R = 0.9Y, p = 0.0002). Matrices subjected to point deformation showed similar response characteristics to those subjected to uniaxial compression, but at a lower level of modulation. When the magnetembedded matrices were exposed to magnetic fields over a range of frequencies available to us (5 to 11 Hz, and 60 Hz) there was a similar linear increase in modulation (Fig. 3; slope 0.12 2 0.02, SEE = 0.80, R = 0.99, p = 0.09). In control experiments neither static compression of the matrices nor DC magnetic field exposure increased release above baseline. Modulation increased with increasing matrix volume during planar stimulation but when the compression was applied to a point on the matrix with an intervening magnetic sphere a parabolic relationship was observed [Fig. 4(a)]. When the influence of matrix radius and matrix height were sepa-
2
6
4
8
MATRIX VOLUME (mm3/100) (4 Figure 4. Thc extent of modulation is plotted against the (a) volume, and (b) radius of the matrix subjected to uniaxial compression (open squares) or point deformation (closed circles) at 60-Hz and 50-pm dynamic compression. Modulation increased linearly with matrix radius (slope 2.61 5 0.45, SEE = 2.20, R = 0.91, p < 0.001) for uniaxial compression. When deformation was applied to a point on the face of the EVAc:BSA matrices through an interposing magnetic sphere placed between the deforming plunger and matrix, a parabolic relationship between matrix geometry and modulation was observed. Error bars represent the standard error for 4 to 16 measurements at each data point.
10
M E C H A N I C A L DEFORMATION OF EVAc: BSA MATRICES
1627
20
15
10
5
n
"0
1
2
3
4
5
6
7
MATRIX RADIUS (mm) (b) Figure 4. (continud)
rated modulation increased linearly with radius [Fig. 4(b), slope 2.61 2 0.45, SEE = 2.20, R = 0.91, p < 0.0011, but not height ( R = 0.09, p = NS), for uniaxial compression. The more complex, parabolic relationship was maintained for point modulation with respect to both matrix height and radius as well [Fig. 4(b)]. Additionally, as the size of the sphere that interceded between the deforming plunger and the matrices was increased, the amount of modulation increased accordingly, and approached values obtained during uniaxial compression [Fig. 51.
DISCUSSION
It has been hypothesized that the effect of magnetic stimulation on the release of drugs from EVAc matrices is the transduction of the applied magnetic field into a mechanical deformation of the matrix through motion of the magnet within the matrix.23 The current study provides support for this hypothesis by obtaining similar modulated release kinetics (Fig. 2) with mechanically compressed and magnetically stimulated matrices. When mechanically or magnetically induced displacements at the same frequency and amplitude were applied to identical matrices the extent of modulation was
EDELMAN ET A L .
3
8E
MAGNET RADIUS (mm) UNIAXIAL Figure 5. The extent of modulated BSA release from EVAc matrices subjected to mechanical deformation (50-pm deformation a t 60 Hz for 5 min) over its entire face or with a n intervening magnetic sphere of 0.7, 1.4, or 3.0 mrn radius is displayed. Modulation increased as more and more of the matrix face was deformed, presumably because more and more of the matrix was subject to modulation.
observed to be, within the same range, and to take the same amount of time to achieve an elevated plateau and return to baseline upon termination of the stimulus (Fig. 2, Table I). The near identity of the time constants of these two systems is in direct contradistinction to modulated release achieved with other types of stimulation. Alteration of porosity of membranes on the basis of local changes in charge or pH have different time constants and probably indicate release mechanisms are governed by different determinants. Ultrasonic stimulation of matrix type release systems yields a similar range of time constants and may bespeak some overlap.’ Uniaxial compression resulted in greater enhancement of release than when a sphere interceded between the plunger and matrix and point deformations were applied. As the size of the sphere was increased, modulations approached those obtained with uniaxial compression (Fig. 5). Modulation increased in a linear fashion with increased deformation frequency for both types of deformation (Fig. 3), but the response to changes in matrix size differed (Fig. 4). Modulation resulting from uniax-
MECHANICAL DEFORMATION OF EVAc:BSA MATRICES
1629
ial compression increased as the matrix size increased and was linear with matrix radius, while point deformation modulation was attenuated at both small and large matrix volumes, radii and heights. A model developed to describe enhanced flow in a two reservoir system connected by a thin channel adequately predicted linear increases in modulation with increasing frequency of stimulation.'' However, the effect of matrix size was not accounted for and there was no provision to handle "point" deformations applied over a limited surface area. This is the type of mechanical effect that might result from an embedded magnet whose surface area was smaller than that of the matrix. The current observations and previous results2,3might be incorporated into a single mechanism by first examining the factors which govern baseline release from the matrices and then the effects that planar and point deformations have on the matrices. Polymeric matrix drug delivery systems are often porous structures which house drug in caves connected by a tortuous network of thin channels. In order for substances that cannot diffuse through the polymer material to be released, they must make their way through this network. This entails the absorption and entry of a fluid to solubilize the dry drug, and the diffusion of solubilized drug through the caves and channels.'-3 One potential explanation for the dependence of the stimulated response on matrix size might be obtained by studying the volume of drug available for release and volume of matrix over which mechanical compression can exert its effect. Assume that in a cylindrical matrix of radius r and height h only a portion of the embedded drug will be solubilized at any point in time. The nonsolubilized portion of the drug phase, that which has remained as a dry powder, is not available for release or modulation. Let us constrain the dry phase to a sphere of radius s residing in the center of the cylindrical matrix. As the matrix increases in size, two competing factors govern release modulated release. First, as the size of the matrix increases, the volume of the solubilized portion of drug increases as m 2 h - 4/37rs3; i.e., the volume of the matrix minus the volume of the sphere of dry drug. Second, the basal rate of diffusion is limited by the surface area of the matrix. In uniaxial compression both the top and bottom faces are in contact with plexiglass and only the sides of the matrix cylinder are available for release. Hence, only 27rrh need be considered. When a magnetic bead is placed between the top of the matrix and the compressing plunger the surface area is increased by ? T ( Y ; ~ , ~~ Y $ ~ ~ If~ the ~ ~magnet ) . radius is much smaller than the matrix face this reduces to mkalrix.Modulation is therefore related to the ratio of the volume available for enhanced release (the volume of solubilized drug) and the surface area available for diffusion. This ratio is:
27rrh mr2h - 4/3rs3 21rrh + m 2
for uniaxial compression, and for point deformation
EDELMAN ET AL.
1630
With time all of the drug is solubilized, the amount of insoluble drug decreases, s approaches 0, and the ratio approaches: r 2
-
rli
(Y
+ 2h)
for uniaxial compression, and for point deformation
Thus, unixial compression as observed in fact is expected to increase with increasing radius as depicted in Figure 4(b). Point deformation is slightly more complicated. Thin, wide matrices (h > r) should demonstrate enhanced release that increases with the radius of the matrix. ,Modulation with point deformation appears to be a more complicated matter; one that did not follow the surface area:volume ratio, and was much more dependent on the existence of an ideal matrix size for optimization of modulation. This may reflect a second competing phenomenon that occurs as the matrix increases in size; the dampening of the force on matrix material distal to the plane or point of force. As the distance from the point of compression increases, the magnitude of the Compression will decay. While the plunger produces a force over the entire face of the matrix, there is only a single point of contact when a sphere intercedes between the plunger and the matrix. Thus, for a uniaxial compression, the amplitude of the displacement of the matrix will decay only in the direction of its application; there will be maximum compression of the matrix in the area immediately below the plunger, and the amplitude of this compression will fall off with the inverse of the distance from the plunger. This is not the case for point deformations where there is a constant decrease in the force exerted on a point in the matrix as one moves radially outward from the point of application. It is possible, therefore, for large matrices to contain a substantial volume of material that is subject to little or none of the deforming force. Drug must still diffuse through this “passive” matrix volume (those portions of the matrix which are not being compressed by the sphere) before being released to the external environment. Thus, reduced release rates and correspondingly lowered modulations should be expected for increased matrix sizes. Similarly, as more of the matrix comes under the influence of the oscillating displacement, modulation should increase. In fact, this was observed, as modulation increased with increasing point of contact (Fig. 5) and with increasing deformation amplitudes. In summary, we have shown that the enhanced release of macromolecules from EVAc matrices that follows the application of external energy fields, such as oscillating magnetic fields and possibly ultrasonic waves, may result from repeated pulsatile mechanical deformation of the matrix. Further studies will hopefully answer some of the questions that remain regarding what properties of the polymer materials, such as elasticity and hydrophobicity, what properties of the drug, such as molecular weight, and what aspects of matrix geometry will be important in determining release characteristics. It is
MECHANICAL DEFORMATION OF EVAc: BSA MATRICES
1631
possible that the same forces that determine modulated release in this system are crucial for control of other forms of modulated release, e.g., with the application of ultrasonic fields to matrix drug release systems.’ The further application of the mechanics of solute transport through porous viscoelastic materials to externally controlled drug delivery devices will hopefully provide the framework for a unified model of the events which lead to enhanced release, and a means of optimizing such systems. This study was supported by a grant from the National Institute of Health, Grant GM44884. Dr. Edelman is currently supported by a Physician-Scientist Award from the National Institutes of Health (K12 AG00294).
References 1. W. Rhine, D. tlsieh, and R. Langer, “Polymers for sustained rnacromolecule release: Procedures to fabricate reproducible delivery systems and control release kinetics,” J. Pharm. Sci., 69, 2 6 5 2 7 0 (1980). 2. E. R. Edelman, J. Kost, H. Bobeck, and R. Langer, “Regulation of drug release from polymer matrices by oscillating magnetic fields,” 1. Hiomed. Mater. Res., 19, 67-83 (1985). 3. E. R. Edelrnan, L. Brown, J. Taylor, and R. Langer, “Regulation of drug release from polymer matrices using oscillating magnetic fields,” I. Biomed. Mater. Res., 21, 339-351 (1987). 4. M. J. McCarthy, D. S. Soong, and E. R. Edelman, “Control of drug release from polymer matrices impregnated with magnetic beads-a proposed mechanism and model for enhanced release,“ ]. Controlled Rel., 1, 143-147 (1984). 5. R. Langer, L. Brown, and E. R. Edelman, “Controlled release and magnetically modulated release systems for macromolecules,” Drug Enzyme Targeting M d h . Enzymol., 112, 399-423 (1985). 6. S.S. Sternstein, “Transient and dynamic characterization of viscoelastic solids,” in Polymer Characterization, Am. Chem. SOC.,Washington, D.C., 1983, pp. 123-147. 7. J. Kost, K. Lcong, and R. Langer, “Ultrasonic modulated drug delivery systems,” in Polymers, MtTdicine, Biomedical and Pharmaceutical Applicntions II, P. Chiellini, P. Gustig, C. Migliaresi, and L. Nicholas (eds.), Plenum Press, New York, 1986, pp. 387-396.
Received December 9, 1991 Accepted April 24, 1992
that repeated pulsatile mechanical deformation of matrices can enhance the release of macromolecules from ethylenevinyl acetate copolymer matrices. Furthermore, similar modulated release kinetics were obtained with mechanically compressed and magnetically stimulated matrices. We also established that modulation was dependent on the ratio of compression area to matrix volume and that modulation was maximized when this ratio was optimized. 0 1492 John Wiley & Sons, Inc.
INTRODUCTION
Controlled release of a variety of compounds has been achieved with polymer-based drug delivery systems.' External modulation of this release has been demonstrated in a number of systems. When magnetic fields were applied to polymer matrices of ethylene-vinyl acetate (EVAc) embedded with drug and a small magnet, drug release was increased up to 30-fold above baseline This phenomenon was hypothesized to involve a mechanical interaction between the applied field and the polymer-drug matrix. Indeed, magnified video images of the surface of an EVAc matrix embedded with a magnet showed alternating compression and decompression of the matrix material adjacent to the magnet during exposure of the matrix to an oscillating magnetic field.3 A preliminary theoretical construct involving mechanical deformation of EVAc-drug matrices followed mass transfer between two solute reservoirs *To whom correspondence should be addressed. Journal of Biomedical Materials Kesearch, Vol. 26, 1619-1631 (1992) CCC 0021-9304/92/121619-13 0 1992 John Wiley & Sons, Inc.
EDELMAN ET AI,.
1620
connected by a long thin tube. This geometry was used to model the drug release from the network of caves and connecting channels that compose the polymer-drug matrices. Solute transfer rate between the reservoirs was obtained by solving the diffusion equation at points far from the ends of the tube, using the velocity profile established by an oscillating piston compressing the end of one tube. The increase in mass transfer above diffusion was found to be a function of three dimensionless groups: an oscillating Reynolds number, an amplitude factor relating the displacement of the piston to the radius of the tube, and the Schmidt n ~ m b e r . ~ In this study we investigated the potential of precisely controlled mechanical compression of EVAc-drug matrices to modulate drug release in practice. This precise control permitted us to investigate whether the same parameters that were important in affecting magnetic controlled modulation played a similar role in modulation obtained with direct mechanical compression and to investigate how the matrix itself might influence the extent of release modulation.
MATERIALS A N D METHODS
Materials Bovine serum albumin (BSA, Sigma Chemical Co., St. Louis, MO, M, 68,800) was sieved to particle sizes between 180 and 250 pm using standard USA testing sieves (ASTME specifications, Dual Manufacturing Co., Chicago, IL). Ethylene-vinyl acetate copolymer (EVAc, 40% vinyl acetate), manufactured under the product name ELVAX-40P, was washed in distilled water and 95% alcohol to remove impurities, and then dissolved in methylene chloride to form a 10% weight by volume s ~ l u t i o n . ' - Magnetized ~.~ stainless steel spheres (80% iron, 17% chromium) were obtained from Ultraspherics Inc. (Ann Arbor, MI). Matrix fabrication Matrices were fabricated in an identical fashion to those used in previous controlled release BSA was added to the EVAc solution to make a 33% weight by volume suspension, vortexed, let stand for 15 s to allow air bubbles to settle out and then poured into glass molds that had been precooled on dry ice. When magnets were embedded within the matrices they were dropped into the hardening EVAc: BSA solution 15 s after pouring. This left the magnets to reside in the center of the matrix. Once hardened, the matrices were removed from their molds, placed in a -20°C freezer for 2 days and then placed under house vacuum (600 mTorr) for another 2 days. The resulting matrix was a homogeneous dispersion of BSA within EVAc. Different sized and shaped glass molds were used to cast cylindrical matrices of varying heights and diameters. When placed in solution predictable and repro-
MECHANICAL DEFORMATIOK O F EVAc:BSA MATIIICES
1621
ducible release kinetics were ob~erved.'-~,~ To assure that we were operating within the linear range of release all matrices were soaked in 0.1M phosphate buffer for 72 h. This allowed for the surface BSA to elute off of the matrix surface and eliminated the burst phase of release. Controlled compression Our model reconstructed the magnetic system by examining the interface between the magnetic spheres and the polymer matrix. In the magnetic system compression occurred within the center of the matrix. As this was inaccessible we applied controlled compression at the surface of the matrices. We hoped to determine whether mechanical compression might produce the same kinetics of release, verifying the validity of the model, and thereafter permit us to examine the importance of various parameters effecting modulatable release. A Dynastat dynamic mechanical spectrometer (IMASS, Inc., Hingham, MA) was used to impose sinusoidal compressive displacements of EVAc matrices over a range of frequencies. The instrument is described in detail elsewhere.(' A special compression chamber was constructed from plexiglass in the shape of a cup with a raised pedestal in the center [Fig. l(a)]. The chamber was 3 cm in diameter and 2.5 cm high and the pedestal, 1.25 cm high and 1.5 cm in diameter. The matrices were placed on the pedestal and release of BSA was allowed to reach to equilibrium after the chamber was filled with 8 mL of 0.1M phosphate buffer at pH 7.4. Compression was applied to the matrix face in one of four different configurations. In the first three, one of different radii spherical magnets (0.7, 1.4, and 3.0 mm) used in the magnetic system was placed between the plunger and the matrix [Fig. l(b)]. In the fourth case the spheres were removed and the samples were then compressed by a solid plexiglass plunger fixed to the upper jaw of the Dynastat. The plunger face was 2 cm in diameter and covered the entire matrix, thus, in the latter case the samples were subjected to uniaxial unconfined compression while completely immersed in buffer. These four configurations modeled the situation wherein magnets of increasing size or of greater ability to affect modulation were used. Mechanical matrix stimulation with both uniaxial compression and point deformation involved a static offset displacement, to keep the matrix in position, and a superimposed adjustable dynamic displacement for potential modulation of BSA release. Magnetic field application Magnetic fields were generated from two different devices.23In the first, permanent magnets were secured to a plate rotating beneath the matrices. Magnetic fields ranged from 0 to 60 Hz at 1100 G peak to peak. In the second an electromagnetic field was used (plate demagnetizer, O.S. Walker Co., Inc., Worcester, M A ) capable of delivering a 60-Hz magnetic field at amplitudes
EDELMAN ET AL.
1622
MATRIX
I
RESTRAINING
CEIl.
-I
-..:
MAGNET
Figure 1. EVAc:BSA matrices were subjected to controlled mechanical deformation. The matrices were placed on a raised pedestal in the center of a plexiglass chamber that could be filled with fluid. RSA was released into the fluid at steady state. As the matrix was deformed by the plunger release rates increased in a manner dictated by the deforming force and frequency. In separate experiments the plunger was either in (a) direct contact with the matrix face or (b) a magnetic bead of variable size was placed in between. The mechanical force in the former experiments was referred to as uniasinl compression and in the latter a s point deformation.
varying from 0 to 900 G (peak to peak). The amplitude of the magnetic field was set to a value capable of displacing the embedded magnet 50 Fm. Release kinetics
Release was followed continuously or at selected intervals. For continuous monitoring of BSA release from the matrices a flow-through spectrophotometer was employed in the open loop c~nfiguration.~ lnlet and outlet ports of the compression chamber allowed for a continuous pumped flow of phosphate buffer (pH 7.4) over the matrix. The outlet solution was pumped through the spectrophotometer, where its absorption was recorded at 280 nm.
MECHANICAL DEFORMATION OF EVAc: BSA MATRICES
1623
The recorded data was then digitized and stored on a computer (SUN Microsystems, Mountain View, CA) for later analysis. System dynamics were defined in the following fashion. Phosphate buffer was pumped through the spectrophotometer cell and at various times BSA solutions of known concentrations were injected into the cell, and the resulting changes in absorbance recorded. Mixing of the BSA solution with the phosphate buffer in the cell and tubing established an exponential rise and decay in absorbance. In subsequent analyses of the matrix response to mechanical or magnetic stimulation the system rise and decay values were used to remove the secondary effects introduced by the flow-through system. This was accomplished by digitally cross-correlating the stimulated response to the system response. In order to follow the BSA release periodically, the solution bathing the matrix was sampled after time intervals of 5 or 10 min. This was accomplished by draining the buffer from the cell and storing it in a glass scintillation vial; the cell was then refilled with 8 mL of fresh buffer. The absorbance at 280 nm was measured to determine the amount of BSA present in each sample. The extent of release enhancement, “modulation,” was defined as the ratio of mass transfer of drug out of the matrix during stimulation to the basal rate of release. This was calculated by dividing the absorbance attained during the application of a dynamic deformation by the baseline absorbance measured in the absence of any oscillating deformation.
RESULTS
System dynamics Solutions of known concentrations of BSA injected through the flowthrough circuit into the compression chamber appeared as a rise in absorbance after 2 min; this delay reflected the transit time through the tubing connecting the chamber and the spectrophotometer. The plateau and baseline BSA concentrations determined from the absorbance recorded in both the flow-through and periodic configurations identically matched the known injected BSA concentration ( p < 0.0001,R = 0.999), indicating no loss of BSA in the sample chamber or circuit tubing. The system response was determined for the controlled injection of five BSA concentrations from 0.1 to 0.5 mg/mL. Matrix deformation and dynamics of release enhancement To determine the validity of our model we examined the BSA release profile from EVAc matrices subjected to mechanical or magnetic stimulation (Fig. 2). A mechanical compressive displacement of 50 pm was applied to the matrix, and compared to the results with application of a 60-Hz magnetic field at an amplitude capable of deforming an identical matrix by displacing the internal magnetic bead by 50 pm. Both curves demonstrate negligible
EDELMAN ET AI..
360
720
1080
1440
Time (seconds) Figure 2. BSA release kinetics from a n EVAc matrices subjected to mechanical deformation were virtually identical to the kinetics that followed magnetic field stimulation of an EVAc: BSA matrix embedded with a magnetic sphere. A 14-min matrix stimulation was applied at 0 time in each case. A 50-pn deformation of the matrix was applied at 60 Hz and compared to a 60-Hz oscillating magnet field capable of displacing the matrix material adjacent to the embedded magnet 50 k m as well. Though different degrees of modulation were observed the amount of BSA released was normalized to allow for direct visual comparison of the kinetics of the two different means of matrix stimulation.
baseline BSA release which increased almost immediately with mechanical deformation or magnetic field exposure. After a stable plateau was obtained and the stimulus was removed, release returned to baseline almost immediately. The rise and decay time constants associated with initiation and termination of matrix stimulation were obtained for both mechanical compression and magnetic field stimulation of EVAc matrices (Table I) and matched identically ( p < 0.0001). Mechanical deformation When the EVAc-BSA matrices were exposed to dynamic uniaxial compression, modulation increased linearly with frequency (5 to 60 Hz, Fig. 3; slope
MECHANICAL DEFORMATION OF EVAc: BSA MATRICES
1625
TABLE I Time Constants of EVAc: BSA Matrix Response Kinetics to Mechanical and Magnetic Stimulation Derived from Curves Depicted i n Figure 2 Time Constant (ms)
Mechanical Magnetic
Kise
Decay
87.41 87.34
71.94 73.53
12
10
4
2' 3
I
10
I
2 0
I
I
I
4 0
50
6 0
1
30
FREQUENCY (Hz) Figure 3. The extent of modulation of BSA release from EVAc matrices is plotted against the frequency of mechanical matrix deformation (closed circles) and magnetic field oscillation (open circles). Identical EVAc : BSA matrices were used. In the magnetic experiment a spherical magnet was embedded within the EVAc:BSA matrix. The matrix was then exposed to a 60-Hz oscillating magnetic field capable of displacing the matrix material around the magnet 50 pm. Mechanical studies were conducted with the magnet placed between the deforming plunger and the matrix; the plunger was set to provide a 50-pm displacement at 60 Hz. A linear increase in modulation was observed for both types of stimulation and within the same range for both magnetic (slope 0.12 ? 0.02, SEE = 0.80, R = 0.99, p = 0.09), and mechanical (slope 0.13 2 0.01, SEE = 0.12, R = 0.99, p = 0.0002) forms of matrix release manipulation.
70
EDELMAN El' AI,.
1626
0.13 % 0.01, SEE = 0.12, R = 0.9Y, p = 0.0002). Matrices subjected to point deformation showed similar response characteristics to those subjected to uniaxial compression, but at a lower level of modulation. When the magnetembedded matrices were exposed to magnetic fields over a range of frequencies available to us (5 to 11 Hz, and 60 Hz) there was a similar linear increase in modulation (Fig. 3; slope 0.12 2 0.02, SEE = 0.80, R = 0.99, p = 0.09). In control experiments neither static compression of the matrices nor DC magnetic field exposure increased release above baseline. Modulation increased with increasing matrix volume during planar stimulation but when the compression was applied to a point on the matrix with an intervening magnetic sphere a parabolic relationship was observed [Fig. 4(a)]. When the influence of matrix radius and matrix height were sepa-
2
6
4
8
MATRIX VOLUME (mm3/100) (4 Figure 4. Thc extent of modulation is plotted against the (a) volume, and (b) radius of the matrix subjected to uniaxial compression (open squares) or point deformation (closed circles) at 60-Hz and 50-pm dynamic compression. Modulation increased linearly with matrix radius (slope 2.61 5 0.45, SEE = 2.20, R = 0.91, p < 0.001) for uniaxial compression. When deformation was applied to a point on the face of the EVAc:BSA matrices through an interposing magnetic sphere placed between the deforming plunger and matrix, a parabolic relationship between matrix geometry and modulation was observed. Error bars represent the standard error for 4 to 16 measurements at each data point.
10
M E C H A N I C A L DEFORMATION OF EVAc: BSA MATRICES
1627
20
15
10
5
n
"0
1
2
3
4
5
6
7
MATRIX RADIUS (mm) (b) Figure 4. (continud)
rated modulation increased linearly with radius [Fig. 4(b), slope 2.61 2 0.45, SEE = 2.20, R = 0.91, p < 0.0011, but not height ( R = 0.09, p = NS), for uniaxial compression. The more complex, parabolic relationship was maintained for point modulation with respect to both matrix height and radius as well [Fig. 4(b)]. Additionally, as the size of the sphere that interceded between the deforming plunger and the matrices was increased, the amount of modulation increased accordingly, and approached values obtained during uniaxial compression [Fig. 51.
DISCUSSION
It has been hypothesized that the effect of magnetic stimulation on the release of drugs from EVAc matrices is the transduction of the applied magnetic field into a mechanical deformation of the matrix through motion of the magnet within the matrix.23 The current study provides support for this hypothesis by obtaining similar modulated release kinetics (Fig. 2) with mechanically compressed and magnetically stimulated matrices. When mechanically or magnetically induced displacements at the same frequency and amplitude were applied to identical matrices the extent of modulation was
EDELMAN ET A L .
3
8E
MAGNET RADIUS (mm) UNIAXIAL Figure 5. The extent of modulated BSA release from EVAc matrices subjected to mechanical deformation (50-pm deformation a t 60 Hz for 5 min) over its entire face or with a n intervening magnetic sphere of 0.7, 1.4, or 3.0 mrn radius is displayed. Modulation increased as more and more of the matrix face was deformed, presumably because more and more of the matrix was subject to modulation.
observed to be, within the same range, and to take the same amount of time to achieve an elevated plateau and return to baseline upon termination of the stimulus (Fig. 2, Table I). The near identity of the time constants of these two systems is in direct contradistinction to modulated release achieved with other types of stimulation. Alteration of porosity of membranes on the basis of local changes in charge or pH have different time constants and probably indicate release mechanisms are governed by different determinants. Ultrasonic stimulation of matrix type release systems yields a similar range of time constants and may bespeak some overlap.’ Uniaxial compression resulted in greater enhancement of release than when a sphere interceded between the plunger and matrix and point deformations were applied. As the size of the sphere was increased, modulations approached those obtained with uniaxial compression (Fig. 5). Modulation increased in a linear fashion with increased deformation frequency for both types of deformation (Fig. 3), but the response to changes in matrix size differed (Fig. 4). Modulation resulting from uniax-
MECHANICAL DEFORMATION OF EVAc:BSA MATRICES
1629
ial compression increased as the matrix size increased and was linear with matrix radius, while point deformation modulation was attenuated at both small and large matrix volumes, radii and heights. A model developed to describe enhanced flow in a two reservoir system connected by a thin channel adequately predicted linear increases in modulation with increasing frequency of stimulation.'' However, the effect of matrix size was not accounted for and there was no provision to handle "point" deformations applied over a limited surface area. This is the type of mechanical effect that might result from an embedded magnet whose surface area was smaller than that of the matrix. The current observations and previous results2,3might be incorporated into a single mechanism by first examining the factors which govern baseline release from the matrices and then the effects that planar and point deformations have on the matrices. Polymeric matrix drug delivery systems are often porous structures which house drug in caves connected by a tortuous network of thin channels. In order for substances that cannot diffuse through the polymer material to be released, they must make their way through this network. This entails the absorption and entry of a fluid to solubilize the dry drug, and the diffusion of solubilized drug through the caves and channels.'-3 One potential explanation for the dependence of the stimulated response on matrix size might be obtained by studying the volume of drug available for release and volume of matrix over which mechanical compression can exert its effect. Assume that in a cylindrical matrix of radius r and height h only a portion of the embedded drug will be solubilized at any point in time. The nonsolubilized portion of the drug phase, that which has remained as a dry powder, is not available for release or modulation. Let us constrain the dry phase to a sphere of radius s residing in the center of the cylindrical matrix. As the matrix increases in size, two competing factors govern release modulated release. First, as the size of the matrix increases, the volume of the solubilized portion of drug increases as m 2 h - 4/37rs3; i.e., the volume of the matrix minus the volume of the sphere of dry drug. Second, the basal rate of diffusion is limited by the surface area of the matrix. In uniaxial compression both the top and bottom faces are in contact with plexiglass and only the sides of the matrix cylinder are available for release. Hence, only 27rrh need be considered. When a magnetic bead is placed between the top of the matrix and the compressing plunger the surface area is increased by ? T ( Y ; ~ , ~~ Y $ ~ ~ If~ the ~ ~magnet ) . radius is much smaller than the matrix face this reduces to mkalrix.Modulation is therefore related to the ratio of the volume available for enhanced release (the volume of solubilized drug) and the surface area available for diffusion. This ratio is:
27rrh mr2h - 4/3rs3 21rrh + m 2
for uniaxial compression, and for point deformation
EDELMAN ET AL.
1630
With time all of the drug is solubilized, the amount of insoluble drug decreases, s approaches 0, and the ratio approaches: r 2
-
rli
(Y
+ 2h)
for uniaxial compression, and for point deformation
Thus, unixial compression as observed in fact is expected to increase with increasing radius as depicted in Figure 4(b). Point deformation is slightly more complicated. Thin, wide matrices (h > r) should demonstrate enhanced release that increases with the radius of the matrix. ,Modulation with point deformation appears to be a more complicated matter; one that did not follow the surface area:volume ratio, and was much more dependent on the existence of an ideal matrix size for optimization of modulation. This may reflect a second competing phenomenon that occurs as the matrix increases in size; the dampening of the force on matrix material distal to the plane or point of force. As the distance from the point of compression increases, the magnitude of the Compression will decay. While the plunger produces a force over the entire face of the matrix, there is only a single point of contact when a sphere intercedes between the plunger and the matrix. Thus, for a uniaxial compression, the amplitude of the displacement of the matrix will decay only in the direction of its application; there will be maximum compression of the matrix in the area immediately below the plunger, and the amplitude of this compression will fall off with the inverse of the distance from the plunger. This is not the case for point deformations where there is a constant decrease in the force exerted on a point in the matrix as one moves radially outward from the point of application. It is possible, therefore, for large matrices to contain a substantial volume of material that is subject to little or none of the deforming force. Drug must still diffuse through this “passive” matrix volume (those portions of the matrix which are not being compressed by the sphere) before being released to the external environment. Thus, reduced release rates and correspondingly lowered modulations should be expected for increased matrix sizes. Similarly, as more of the matrix comes under the influence of the oscillating displacement, modulation should increase. In fact, this was observed, as modulation increased with increasing point of contact (Fig. 5) and with increasing deformation amplitudes. In summary, we have shown that the enhanced release of macromolecules from EVAc matrices that follows the application of external energy fields, such as oscillating magnetic fields and possibly ultrasonic waves, may result from repeated pulsatile mechanical deformation of the matrix. Further studies will hopefully answer some of the questions that remain regarding what properties of the polymer materials, such as elasticity and hydrophobicity, what properties of the drug, such as molecular weight, and what aspects of matrix geometry will be important in determining release characteristics. It is
MECHANICAL DEFORMATION OF EVAc: BSA MATRICES
1631
possible that the same forces that determine modulated release in this system are crucial for control of other forms of modulated release, e.g., with the application of ultrasonic fields to matrix drug release systems.’ The further application of the mechanics of solute transport through porous viscoelastic materials to externally controlled drug delivery devices will hopefully provide the framework for a unified model of the events which lead to enhanced release, and a means of optimizing such systems. This study was supported by a grant from the National Institute of Health, Grant GM44884. Dr. Edelman is currently supported by a Physician-Scientist Award from the National Institutes of Health (K12 AG00294).
References 1. W. Rhine, D. tlsieh, and R. Langer, “Polymers for sustained rnacromolecule release: Procedures to fabricate reproducible delivery systems and control release kinetics,” J. Pharm. Sci., 69, 2 6 5 2 7 0 (1980). 2. E. R. Edelman, J. Kost, H. Bobeck, and R. Langer, “Regulation of drug release from polymer matrices by oscillating magnetic fields,” 1. Hiomed. Mater. Res., 19, 67-83 (1985). 3. E. R. Edelrnan, L. Brown, J. Taylor, and R. Langer, “Regulation of drug release from polymer matrices using oscillating magnetic fields,” I. Biomed. Mater. Res., 21, 339-351 (1987). 4. M. J. McCarthy, D. S. Soong, and E. R. Edelman, “Control of drug release from polymer matrices impregnated with magnetic beads-a proposed mechanism and model for enhanced release,“ ]. Controlled Rel., 1, 143-147 (1984). 5. R. Langer, L. Brown, and E. R. Edelman, “Controlled release and magnetically modulated release systems for macromolecules,” Drug Enzyme Targeting M d h . Enzymol., 112, 399-423 (1985). 6. S.S. Sternstein, “Transient and dynamic characterization of viscoelastic solids,” in Polymer Characterization, Am. Chem. SOC.,Washington, D.C., 1983, pp. 123-147. 7. J. Kost, K. Lcong, and R. Langer, “Ultrasonic modulated drug delivery systems,” in Polymers, MtTdicine, Biomedical and Pharmaceutical Applicntions II, P. Chiellini, P. Gustig, C. Migliaresi, and L. Nicholas (eds.), Plenum Press, New York, 1986, pp. 387-396.
Received December 9, 1991 Accepted April 24, 1992
