Erosion kinetics of hydrolytically degradable polymers

Transcription

1 Proc. Nadl. Acad. Sci. USA Vol. 9, pp , January 1993 Chemistry Erosion kinetics of hydrolytically degradable polymers J. A. TAMADA* AND R. LANGERt Massachusetts Institute of Technology, Department of Chemical Engineering, Cambridge, MA 2139 Contributed by R. Langer, September 25, 1992 ABSTRACT Degradable polymers are beginning to play an increasing role as materials for environmental and medical applications. Understanding factors that control erosion, such as bond cleavage and the dissolution and diffusion of degradation products, will be critical to the future development of these materials. Erosion kinetics, photomicroscopy, and infrared spectroscopy were used to understand the erosion mechanism of two families of degradable polymers, polyanhydrides and polyesters. Polyanhydrides exhibit behavior more characteristic of surface erosion, whereas the polyesters exhibit bulk erosion patterns. Control of erosion times from a few days to several years can be achieved by ajudicious choice of monomer units and bond selection. The use of degradable polymers in environmental, medical, and other applications is gaining increased importance. Degradable polymers have potential application as materials for bottles, diapers, and other widely used consumer products that would not create environmental build-up, as well as biomedical applications, including drug delivery systems (1), resorbable sutures (2), stents (3), and scaffolds for tissue regeneration (4). Despite the growing use of these materials, there is only limited understanding of the processes involved in their degradation and erosion kinetics. A greater comprehension of these mechanisms would help optimize current usage and aid in future development of additional degradable materials. The term "degradable" has been applied to polymers that disintegrate by a number of processes, including physical disintegration, biodegradation by biological mechanisms, and chemical reaction to monomer units. Materials that undergo only physical disintegration have limited utility. The material becomes less visible, but the same quantity and composition of the material persists. Breakdown by biodegradation is difficult to predict and control, because of high variability in biologically mediated processes. Degradation by chemical reaction to soluble monomer units might be a more controllable and practical alternative. Moreover, if the monomers are metabolized by biological systems, the material completely disappears from its surroundings. In this report, two major classes of degradable polymer systems, polyanhydrides and polyesters, that undergo degradation by chemical reaction are examined. Approaches for analysis that apply to erodible polymer systems in general are presented. The polyanhydrides studied are copolymers of 1,3-bis(p-carboxyphenoxy)propane (CPP) and sebacic acid (SA), a polymer system currently in clinical trials for delivering drugs for treating brain cancer (5). The polyester poly(lactic/glycolic acid) [p(lga)], a copolymer of lactic acid (LA) and glycolic acid (GA), is from a family of Food and Drug Administration-approved polymers that are used in resorbable sutures (2, 6) and injectable drug delivery systems (7). Both classes of polymers degrade into nontoxic products (8, 9). The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C solely to indicate this fact. 552 Both polyesters and polyanhydrides undergo hydrolytic bond cleavage to form water-soluble degradation products that can dissolve in an aqueous environment, resulting in polymer erosion. In this context, the term "degradation" refers to the bond cleavage reaction, whereas "erosion" refers to the depletion of material. Degradation is a chemical phenomenon; erosion encompasses physical phenomena, such as dissolution and diffusion. Polymer erosion can be ideally divided into "bulk erosion" and "surface erosion." For ideal bulk erosion, material is lost from the entire polymer volume. The erosion rate depends on the total amount of material and generally decreases as material is depleted. The length of time the polymer persists can be altered by changes in chemical composition but not by the material's size or shape. In surface erosion, material is lost from the polymer matrix exterior surface. For ideal surface erosion, erosion rate is directly proportional to external surface area. Thus, for a thin flat slab, for which the external surface area remains constant as the slab becomes progressively thinner, the erosion rate is essentially constant until the polymer is completely eroded. For a surface-eroding polymer, control of the time span the polymer persists can be achieved by adjusting the material's dimensions and shape and by changing its chemical properties. For many applications, bulk erosion is satisfactory. For example, for degradable plastics for packaging, any erosion mechanism is satisfactory so long as the material maintains its integrity during use and disappears completely after disposal. However, for other applications, surface erosion is more desirable or possibly essential. For example, in drug delivery applications, a surface-eroding polymer can provide constant and easily controllable drug release rates and protect the drug from in vivo degradation, whereas bulk erosion may give decreasing release rates and provides less protection for the drug from body fluids. Most polymers display bulk erosion characteristics, as surface erosion is difficult to achieve. For most polymers, water penetrates and degrades the interior of the material faster than the surface can erode. The materials studied are made of two types of monomers. This copolymer approach is attractive because it imparts the capability to tailor the polymer's properties by simply changing the monomer ratio. For bioerodible materials, one property of major interest is erosion rate. For the p(cpp-sa) polyanhydrides, as the CPP content is increased, there is a monotonic decrease in erosion rates. Erosion periods from a few days for p(sa), to 2 years for p(cpp), are achievable for 1-mm-thick polyanhydride disks (1). For p(lga) copolymers, the erosion rate is maximal at an =5:5 mol % LA to GA ratio; the homopolymers p(la) and p(ga) erode more slowly. The effect of composition on erosion rate is generally attributed to the lower crystallinity of the 5:5 mol % Abbreviations: CPP, 1,3-bis(p-carboxyphenoxy)propane; SA, sebacic acid; p(lga), poly(lactic/glycolic acid); LA, lactic acid; GA, glycolic acid; p(sa), p(cpp), etc., poly(sa), poly(cpp), etc. *Present address: Cygnus Therapeutic Systems, Redwood City, CA tto whom reprint requests should be addressed.

2 Chemistry: Tamada and Langer copolymer allowing greater absorption of water into the polymer and faster degradation and erosion (7). MATERIALS AND METHODS Polymers. Phosphate buffer and methylene chloride were from Mallinckrodt. All other inorganic reagents were from Aldrich. The polyanhydrides p(sa) and p(cpp) were synthesized according to ref. 1 and provided by Nova Pharmaceuticals (Baltimore). p(sa) disks and p(cpp-sa) were compressionmolded (1) from <25-mm ground powder at room temperature. The p(sa) and p(cpp-sa) were spray-dried into a fine powder disk in a Buchi model 19 spray dryer with methylene chloride as the solvent. The disk thicknesses were determined by the average of at least three micrometer readings of the dry device. For studies involving p(sa), SA content in supernatants was measured by HPLC on a Hamilton PRP-1 reverse-phase column (Rainin, Woburn, MA) with 15% (vol/ vol) acetonitrile/85% (vol/vol).1 M tetrabutylammonium phosphate (PIC A, Waters) in water, a flow rate of 1.2 ml/min, UV detection at 21 nm, and a 1-AJ injection volume. Polymer molecular weights were determined relative to polystyrene standards (Polysciences) by gel permeation chromatography on a Perkin-Elmer GPC system consisting of a series 1 pump and an LKB 214 rapid spectral detector at 254 nm. Samples were eluted with chloroform through a 3 x.75 cm PL gel column with a 5-mm particle size (Polymer Laboratories, Amherst, MA). For studies involving p(cpp- SA), SA and CPP content were measured by HPLC as above, except for a gradient mobile phase composition. The initial mobile-phase composition was acetonitrile/.5 M tetrabutylammonium phosphate, 16:84 (vol/vol), in water. The initial ratio was maintained for 6 min to allow elution of SA, linearly ramped to 22:78 (vol/vol) over 2-3 min, then to 28:72 (vol/vol) over 2-3 min, maintained at 28:72 (vol/vol) for 5 min, ramped to 16:84 (vol/vol) over 3 sec, and maintained at 16:84 (vol/vol) for 8 min to allow column equilibration. p(lga) 5:5 mol % (Mr, 94,5) was obtained from Medisorb (Cincinnati). Disks (14 mm diameter) were prepared by melt molding at 8 C in Teflon molds. LA content was determined by HPLC using a 15-cm-long 4.1-mm i.d. Hamilton PRP-X3, sulfonic acid functional groups, mobile phase of H2SO4 at ph 2., a flow rate of 1. ml/min, 2-ml injection volume, and UV detection at 21 nm. GA was measured by a chromotropic acid assay in which 1 ml of aqueous sample was mixed with 2 ml of 4,5-dihydroxy-2,7- naphthalene disulfonic acid disodium salt dihydrate at 1. mg/ml (Aldrich) in concentrated sulfuric acid, heated 3 min at 1 C, combined with 4 ml of distilled water, and measured by UV absorbance at 578 nm in quartz cuvettes. All erosion studies were performed in.1 M aqueous phosphate buffer (ph 7.4) at 37 C with 11 rpm agitation using a clinical rotator (Thomas Scientific). Supernatants were changed periodically to ensure perfect sink conditions. Visualization Studies. p(cpp-sa) (2:8 mol %; initial weight, 2 mg; initial width, 1.3 mm) and p(lga) [25:75 mol % (Wako Biochemicals, Osaka; Mr, 2,); initial weight, 2 mg; initial width, 1.1 mm] were used. Disks were immersed in phosphate buffer at 37 C to which.1% of the dye acid orange 8 (Aldrich) was added. Disks were removed and cross-sectioned with a razor blade. Micrographs were made on a Wild M42 Makroskop (Heerbrugg, Switzerland), at x 32.5 magnification, with Polaroid 58 instant sheet film. The dye penetrates highly porous regions but not uneroded polymer. Erosion-front advancement was determined from the ratio of intact zone thickness to initial device thickness, as determined from photomicrographs. For compositional studies of the outer and inner zone, the outer zone was scraped Proc. Natl. Acad. Sci. USA 9 (1993) 553 from the inner zone, and both zones were placed into fresh phosphate buffer (ph 7.4), heated, and stirred to allow for complete dissolution. The dissolved samples were analyzed for SA and CPP by HPLC. Infrared Spectroscopy. Infrared spectra of p(cpp-sa) 2:8 mol % were obtained in KBr pellets on a Perkin-Elmer Spectrophotometer model 143. RESULTS AND DISCUSSION Erosion Kinetics. Fig. 1 shows the rate of SA monomer erosion from p(sa) homopolymer disks of various thicknesses and volumes but approximately identical diameters and external surface areas. Initially, disks of all thicknesses show identical erosion rate profiles, with an induction period followed by a peak in erosion rate. Later, erosion rate decreases, and a sharp distinction between the disks of different thicknesses develops. The thicker disks show prolonged erosion at the peak rate, with the length of time before the decreasing rate period in nearly direct proportion to disk thickness. The effect of geometry is similar to what would be expected for surface erosion-disks of the same surface areas give the same rates, but thicker disks erode over a longer time period. In Fig. 2, for p(cpp-sa) 2:8 and 5:5 mol % copolymers and for both SA and CPP monomers, the initial and maximum rates of erosion are nearly identical for disks of different thicknesses; disks that are twice as thick take nearly twice as long to erode-behavior characteristic of surface erosion. The erosion kinetics of p(lga) are a sharp contrast to that of p(cpp-sa). Fig. 2 also shows the erosion kinetics of p(lga) 5:5 mol % disks for two thicknesses. The thin and thick disks take the same amount of time to erode, and the thicker disks give approximately twice the erosion rate for both GA and LA than the thinner disks. Thus for p(lga) it is not the surface area but the volume of the disk that controls the rate of erosion-behavior characteristic of bulk erosion. This kinetic analysis supports previous observations concerning physical changes and water uptake (7) that suggest that p(lga) undergoes bulk erosion. This result is also supported by release data from p(lga) films.4 and.8 mm thick that show nearly identical cumulative release profiles on a percentage basis for the two thicknesses (11). Another feature observed from the polyanhydride erosion profiles is that SA, the more hydrophilic monomer, is eroded more quickly than CPP. The percentage of original CPP content remaining when the SA has been completely depleted increases with the amount ofcpp in the original composition, 1%o for the 2:8 mol % composition and 5% for 5:5 mol %. Similarly, for p(lga), in previous studies (12) and the current study, the more hydrophobic LA monomer persists 4 3 I E FIG. 1. Rate of erosion of p(sa) (Mr, 9) disks. Mass and thickness of disks: o, mg and.38 mm; e, 97. mg and.6 mm; o, 15.6 mg and.94 mm; *, 2. mg and 1.19 mm; A, mg and 1.47 mm.

3 554 Chemistry: Tamada and Langer A 3 Proc. Natl. Acad. Sci. USA 9 (1993) C 1- CD c 1.' E.8 I.6,.4 I Time, days FIG. 2. Rate of erosion of SA and CPP from p(cpp-sa) copolymer p(lga) disks. (A) p(cpp-sa) 2:8 mol % copolymer (M, 88). Data represent an average of three measurements; error bars are standard deviations. Mass and thickness of the disks: *, 1 ± 5 mg and.75 ±.5 mm; o, 2 ± 5 mg and 1.3 ±.3 mm. (B) p(cpp-sa) 5:5 mol % copolymer (Mr 6). Data with error bars represent an average of three measurements; error bars are standard deviations. Mass and thickness of the disks: A, 95 ± 5 mg and.7 mm; *, 1%.2 mg and 1.27 mm. (C) p(lga) 5:5 mol % copolymer (M, 94,5) disks. Data represent an average of two measurements. Mass and thickness of the disks: *, 1 mg and.62 mm; o, 2 mg and 1.18 mm. longer than GA. In our studies, material balances comparing the LA and GA amounts in the original polymer and the cumulative amounts in the release solutions indicated that almost all (>9%) of the initial GA content could be accounted for in release solutions at the study's end, whereas only about half of the initial LA was recovered. Solid remnants of the devices, which presumably contained the remaining LA, persisted at the study's end but were too fragile to sample to obtain a conclusive LA material balance. Physical Changes. Fig. 3 A-D shows cross-sections of p(cpp-sa) 2:8 mol % polyanhydride disks at various times during erosion. The 5:5 mol % copolymer shows similar behavior (data not shown). The orange color highlights an "erosion zone," a layer of essentially degraded highly porous material that expands steadily from the outside to the inside of the disk (12). The inner zone retains the intact appearance of the initial polymer. The outer zone contains <5% of the original material for p(cpp-sa) 2:8 mol % and <3%o for p(cpp-sa) 5:5 mol %. These percentages are estimated from the width of the erosion zone from photomicrographs (Fig. 3) and the amount of material lost as determined from HPLC analysis, assuming the inner zone retained most of its initial mass. The polyanhydrides thus show material being successively exhausted from the outside to the inside, similar to the behavior of a surface-eroding polymer. The steady advancement of the erosion zone is consistent with results from erosion kinetics. Initially, erosion takes place from the exterior of the disk; therefore, disks of different thicknesses and volumes but the same external surface area yield the same erosion kinetics. In Figs. 1 and 2 during the early stages of erosion, for all polyanhydride compositions, disks of different thicknesses but essentially identical external surface areas give essentially identical rates and erosion patterns. Later in the erosion, the front reaches the center of the thinner disks, and different thicknesses yield different rates. The thin disk displays a rate decrease when the erosion front reaches the center because material in the disk is exhausted. At the same time point, the thicker disks still have an intact interior zone and display the high erosion rate because erosion is still occurring from a region with high material content. Because the velocity of erosion front advancement is steady, the erosion zone takes nearly twice as long to reach the center of disks that are twice as thick. In contrast, no erosion zone is observed in cross-sections of the p(lga) 25:75 mol % copolymer (Fig. 3 E-;H). [The polymer in the photographs had a lower molecular weight (2,) than that used in the kinetic studies and degraded more rapidly, so that shorter time points were used.] The polymer is initially brittle and transparent, but as erosion proceeds the polymer matrix becomes pliable and deforms, losing its initial shape. There is no apparent erosion zone as was observed for polyanhydrides. The orange dye is absorbed throughout the polymer. Electron micrographs by previous workers (11, 13) indicate that p(lga) copolymers show pore formation uniformly throughout the matrix. The pores get progressively larger throughout the entire disk as it erodes, with no apparent progression from exterior to interior. This behavior is consistent with the bulk erosion kinetics described above. Because material is released uniformly from the entire volume, a disk with twice the thickness and volume produces monomers at twice the rate as the thinner disk but takes the same total time to erode. Further examinations of the nature of the polyanhydride erosion zone were performed. From micrographs, the progression of the erosion zone edge through the 2:8 and 5:5 mol % copolymer was determined (Fig. 4A). The erosion zone advances steadily toward the disk center. Also plotted is the cumulative fraction of SA appearing in the solution as determined from kinetic studies. There is a close correlation between SA loss from the copolymer and erosion zone progression; SA loss follows erosion zone advancement. In Fig. 4B, the relative SA to CPP composition of the outer and inner zones at various time points during erosion of p(cpp- SA) 2:8 mol % is shown. The outer zone shows a rapid decrease from the initial SA content relative to CPP, whereas the inner zone retains the initial composition for an extended period. From these results, it is concluded that SA is depleted primarily from the erosion zone as it advances from outside to inside, leaving a shell of primarily CPP-containing material. Chemical Changes. The results from the erosion kinetics demonstrated that, for both polyester and polyanhydride 42

4 Chemistry: Tamada and Langer Proc. Natl. Acad. Sci. USA 9 (1993) 555 I6 -% B.8 It.6 INNER.4 ~.2 OUTER g FIG. 3. Cross sections of p(cpp-sa) (Mr, 88) 2:8 mol %; initial weight, 2 mg; initial width, 1.3 mm. Time: A, 6 h; B, 24 h; C, 48 h; D, 72 h. Cross sections of p(lga) 25:75 mol % (Mr, 2,); initial weight, 2 mg; initial width, 1.1 mm. Time: E, h; F, 24 h; G, 72 h; H, 12 h. copolymers, there is more rapid release of the more hydrophilic comonomer. Two contrasting mechanisms for the different erosion rates can be hypothesized, one with degradation of the linkage being the rate limiting step and the other with the dissolution or diffusion of degradation products :X FIG. 4. (A) Advancement of erosion front through 2 mg of p(cpp-sa) 2:8 mol % (A) and 25 mg of p(cpp-sa) 5:5 mol % (o). Lines show cumulative erosion of SA monomer as measured by HPLC. Dashed lines, 2:8 mol %; solid lines, 5:5 mol %. (B) Composition of the outer and inner zones of p(cpp-sa) 2:8 mol % during erosion. being the rate limiting step. Copolymers of monomers A and B have three types of possible linkages, A-A, A-B, and B-B. In the first hypothesis, the A-A, A-B, and B-B linkages have different reactivities and the monomers are liberated at different rates. In the second hypothesis, the degradation rates of the three bond types are identical, and the slow dissolution or diffusion of the more hydrophobic monomer causes slower erosion. Studies were performed to understand the different reactivities of the SA-SA, SA-CPP, and CPP-CPP bonds for polyanhydrides. Infrared spectroscopy was used to determine the relative amount of the anhydride linkage to the carboxyl degradation product in both the inner and outer zones. Fig. 5 shows the infrared spectra of material sampled from p(cpp-sa) 2:8 mol % copolymer disks at various stages of erosion. The two peaks at 181 and 174 cm-' are attributed to the anhydride bond of undegraded polymer; the peak at 16 cm-' is characteristic of the C=C of the aromatic rings of CPP; the peak at 167 cm-' is characteristic of the carboxyl groups of the degradation products. At later stages in erosion, especially for the outer zone, two separate bands at 181 and 177 cm-' are observed. The peak at 177 cm-1 is characteristic of the CPP-CPP anhydride bond and the peak at 181 cm-1 is characteristic of both SA-SA and SA-CPP anhydride linkages. Assignments were given based on homopolymer spectra. As the time course of erosion progresses (Fig. 5) the following changes in the spectral pattern emerge: In the inner zone, at time zero, the peaks of the initial anhydride bonds are present. At 8 h, a peak characteristic of a carboxylic acid degradation product appears. The relative intensity of the carboxyl degradation

5 556 Chemistry: Tamada and Langer p(cpp-sa) 2:8 8 H 2 H 36 H 48 H 6 H No Inner zone I Wave number, cm-' H 2 H 36 H 48 H 6 H 84 H CPP monomer FIG. 5. Infrared spectra of outer (Right) and inner (Left) zones of p(cpp-sa) 2:8 mol %. product peak to the initial anhydride peak increases at later times, revealing that progressive degradation of the anhydride bonds occurs in the intact inner zone. This finding is consistent with previous studies (14) that demonstrated loss of molecular weight in the inner intact zone within hours after immersion in buffer. In the outer zone, characteristic peaks ofthe anhydride and carboxylic acid groups of partially degraded polymer are present at 8 h (Fig. 5). The outer zone contains a relatively higher ratio of degraded carboxyl groups to undegraded anhydride bonds than the inner zone of the same disk. This is consistent with the expectation that degradation occurs more rapidly in the exterior of a polymer matrix that displays surface-eroding characteristics. As erosion proceeds further, the anhydride band characteristic of the CPP-CPP anhydride bond increases in intensity relative to the SA-SA and SA- CPP anhydride band. The anhydride bonds connected to SA are successively broken, whereas CPP-CPP anhydride bonds remain uncleaved. At 84 h, all SA is gone from the sample, all of the anhydride bonds connected to SA have been cleaved, the eroded zone has penetrated the entire volume of Proc. Natl. Acad. Sci. USA 9 (1993) the disk, and there is no inner zone. Yet characteristic bands of CPP-CPP anhydride bonds are still apparent in the spectra. Clearly, the different types of anhydride linkages react at different rates. It can be concluded that differences in the degradation rate of the anhydride bonds connected to SA and CPP affect the erosion of polyanhydride copolymers. Erosion occurs in a moving front from the outside to the inside of the polymer matrix, and SA and CPP that is connected to SA is depleted from the moving front. A skeleton of material, which contains slowly degrading CPP-CPP anhydride bonds, remains uneroded. A polymer with higher CPP content will contain more CPP-CPP bonds and will, therefore, contain more CPP upon SA depletion, as was observed above. In summary, the p(cpp-sa) polyanhydrides display many surface erosion characteristics. (It should be noted that factors other than the polymer's chemical structure, such as polymer matrix morphology or the addition ofa drug or other substance, may also effect the erosion pattern.) These surface erosion characteristics include a pattern of material loss from outside to inside, rate dependence on the surface area rather than the volume of the polymer matrix, and longer lifetimes for thicker samples. The copolymer of p(lga) shows bulk erosion characteristics, with material loss uniformly throughout the polymer matrix, rate dependence on polymer matrix volume, and the same lifetimes for different thickness disks. The approaches discussed above may be helpful in elucidating the degradation behavior of, and possibly designing, additional polymers for a potentially wide variety of medical, consumer, and other applications. This study was supported by National Institutes of Health Grants GM44884 and CA Heller, J. (1984) in Medical Applications ofcontrolled Release, eds. Langer, R. S. & Wise, D. L. (CRC, Boca Raton, FL), Vol. I, pp Chu, C. C. (1985) in CRC Crit. Rev. Biocompatibil. 1, Agrawal, C. M., Haas, K. F., Leopold, D. A. & Clark, H. G. (1992) Biomaterials 13, Cima, L. G., Ingber, D. E., Vacanti, J. P. & Langer, R. (1991) Biotechnol. Bioeng. 38, Brem, H., Mahaley, M. S., Jr., Vick, N. A., Black, K. L., Schold, S. C., Jr., Burger, P. C., Friedman, A. H., Ciric, I. S., Eller, T. W., Cozzens, J. W. & Kenealy, J. N. (1991) J. Neurosurg. 74, Mukherjee, D. P. (1989) in Sutures: Encyclopedia of Polymer Science and Engineering (Wiley, New York), Vol Holland, S. J., Tighe, B. J. & Gould, P. L. (1986) J. Controlled Release 4, Leong, K. W., D'Amore, P., Marletta, M. & Langer, R. (1986) J. Biomed. Mater. Res. 2, Ratcliffe, J. H., Hunneyball, I. M., Smith, A., Wilson, C. G. & Davis, S. S. (1984) J. Pharm. Pharmacol. 36, Leong, K. W., Brott, B. C. & Langer, R. (1985) J. Biomed. Mater. Res. 19, Shah, S. S., Cha, Y. & Pitt, C. G. (1992) J. Controlled Release 18, Wang, H. T., Palmer, H., Linhardt, R. J., Flanagan, D. R. & Schmitt, E. (199) Biomaterials 11, Cohen, S., Yoshioka, T., Lucarelli, M., Hwang, L. H. & Langer, R. (1991) Pharm. Res. 8, D'Emanuele, A., Hill, J., Tamada, J. A., Domb, A. J. & Langer, R. (1992) Pharm. Res. 9, Mathiowitz, E. & Langer, R. (1987) J. Controlled Release 5,