Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

An intratumoral controlled release formulation of clusterin antisense oligonucleotide and paclitaxel… Springate, Christopher Michael Kevin 1999

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-ubc_1999-0382.pdf [ 5.88MB ]
Metadata
JSON: 831-1.0089048.json
JSON-LD: 831-1.0089048-ld.json
RDF/XML (Pretty): 831-1.0089048-rdf.xml
RDF/JSON: 831-1.0089048-rdf.json
Turtle: 831-1.0089048-turtle.txt
N-Triples: 831-1.0089048-rdf-ntriples.txt
Original Record: 831-1.0089048-source.json
Full Text
831-1.0089048-fulltext.txt
Citation
831-1.0089048.ris

Full Text

THE PHYSICOCHEMICAL CHARACTERIZATION OF PACLITAXEL LOADED POLY(ETHYLENE-CO-VINYL ACETATE) AND POLYURETHANE MATRICES FOR PERIVASCULAR APPLICATION FOR THE INHIBITION OF RESTENOSIS by CHRISTOPHER MICHAEL KEVIN SPRINGATE B.Sc. (Pharm), The University of British Columbia R.Ph., The College of Pharmacists of British Columbia A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE in THE FACULTY OF GRADUATE STUDIES Faculty of Pharmaceutical Sciences (Division of Pharmaceutics and Biopharmaceutics) We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA May 1999 © Christopher Michael Kevin Springate, 1999 06/08/99 14:47 FAX 604 822 9587 SPECIAL COLLECTIONS IS002 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, 1 agree that the c Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Department of <m \v(Sfcr4 01= ^&sj^^rfXc^^ TKe University of British Columbia Vancouver, Canada Date c^AA^Jf On /Wf.9 DE-6 (2/88) © ABSTRACT Surgical procedures performed on blood vessels may cause irritation of the vessel wall. This irritation sets off a series of complicated steps, which results in a growth of the vessel wall towards the center of the blood vessel, and a narrowing of the amount of space for the blood to flow through. This process is called restenosis. Restenosis of the treated artery is a major complication of percutaneous transluminal coronary angioplasty (PTCA) and coronary artery bypass grafting. This intimal hyperplasia (thickening of the intimal layer of the blood vessel) is due to the differentiation, migration, and proliferation of connective tissue and vascular smooth muscle cells at the site of vascular injury (Mishaly, 1997). Through this process the arterial lumen may be enclosed by the neointima and thus compromise coronary blood flow. Paclitaxel stabilizes microtubules and in this way inhibits differentiation of vascular smooth muscle cells from a contractile phenotype to a migratory and proliferative phenotype in vitro. In vivo paclitaxel has inhibited vascular smooth muscle cell migration and proliferation in rabbits and in the rat carotid artery model. Drug delivery to the adventitia of the blood vessel may target events occurring in the adventitia, media, and intima. Perivascular drug delivery devices have delivered drugs that reduce stenosis in arteries in animal models. The objective of this work was to develop a flexible, biocompatible, paclitaxel loaded polymer film for perivascular application, to provide controlled release of paclitaxel over several weeks. Ill Poly(ethylene-co-vinyl acetate) (EVA) with monomer ratios of 60/40 and 72/28, and polyurethane, were cast into films with various loadings of paclitaxel. These polymers were chosen because of their biocompatibility, hydrophobicity, flexibility, and because they are nondegradable. Perivascular films were manufactured from 60/40 EVA, 72/28 EVA or polyurethane, with various loadings of paclitaxel and sterilized by y-irradiation. The physicochemical properties, and diffusion and release characteristics of paclitaxel in these films were investigated in this work. Paclitaxel existed within the EVA matrices as a granular, amorphous solid, whereas paclitaxel was miscible with the polyurethane matrices, y-irradiation of paclitaxel loaded EVA films resulted in cross-linking of the polymer chains, whereas y-irradiation of polyurethane films resulted in polymer chain scission. Partitioning, permeability, and diffusion coefficients of paclitaxel in EVA and polyurethane were determined, and were similar for the two different types of matrices. Paclitaxel release from EVA and polyurethane films was linear with the square root of time, and with the square root of the loading concentration, for the first several days. Paclitaxel release from EVA and polyurethane films was by diffusion without the creation of channels or pores, and followed the Higuchi model of release for the first several days. Paclitaxel release from 60/40 EVA, 72/28 EVA, and polyurethane, was influenced by polymer monomer ratio, polymer type, and drug loading. Given the effects of sterilization on paclitaxel loaded 60/40 EVA, 72/28 EVA, and polyurethane films, polyurethane films showed the most promising potential for developing a film for the controlled release of paclitaxel for perivascular application for the inhibition of restenosis. iV TABLE OF CONTENTS ABSTRACT ii TABLE OF CONTENTS iv LIST OF TABLES k LIST OF FIGURES X LIST OF ABBREVIATIONS xiii ACKNOWLEDGEMENTS x x 1. INTRODUCTION 1 2. BACKGROUND 6 2.1. Restenosis 6 2.1.1. Anatomy of arterial wall 6 2.1.2. Mechanisms of restenosis 8 2.1.3. Pharmacological interventions and drug delivery systems for restenosis 11 2.2. Paclitaxel 13 2.2.1.1. Rationale for selection of paclitaxel as an inhibitor of restenosis .13 2.2.1.2. Mechanism of action of paclitaxel as an anti-restenotic agent 14 2.2.1.3. Toxicities of paclitaxel 15 2.2.1.4. Pharmacokinetics of paclitaxel 16 2.2.2. Physicochemical properties of paclitaxel 17 2.3. Polymers 18 2.3.1. Morphology of polymers 18 2.3.1.1. Polymer constitution 18 2.3.1.2. Polymer configuration 20 2.3.1.3. Polymer conformation 20 2.3.2. Polymer molecular weight 21 2.3.2.1. Definitions of polymer molecular weights 21 2.3.2.2. Measurement of polymer molecular weight 22 2.3.2.2.1. Viscometry of polymers 22 2.3.2.2.2. Gel permeation chromatography (GPC) 23 2.3.3. Polymer crystallinity and crystallization 24 2.3.3.1. Polymer crystallinity 24 2.3.3.2. Measurement of polymer crystallinity 26 2.3.4. Glass transition temperature (Tg) of polymers 27 2.3.4.1. Factors affecting Tg of polymers 27 2.3.5. Mechanical properties of polymers 29 2.3.5.1. Types of mechanical deformations of polymers 29 2.3.5.1.1. Simple extension (tensile) deformation of polymers 29 2.3.5.1.2. Simple shear deformation of polymers 30 2.3.5.1.3. Bulk compression deformation of polymers 31 2.3.5.2. Viscoelasticity of polymers 31 2.3.5.2.1. Definition of elastic body 32 2.3.5.2.2. Definition of viscous liquid 33 2.3.5.2.3. Definition of viscoelastic fluid 33 2.3.5.2.4. Regions of viscoelasticity of polymers 33 2.3.5.3. Dynamic mechanical analysis (DMA) of polymers 35 2.4. Polymeric drug delivery 38 2.4.1. Rationale for the use of a polymer drug delivery system for the controlled, local delivery of an inhibitor of restenosis 38 2.4.2. Drug release from non-porous, non-degradable polymer matrices 39 2.4.2.1. Drug diffusion through polymer matrices 39 2.4.2.2. Drug release from non-porous non-degradable polymer matrices 44 2.4.3. Variables affecting drug diffusion through and drug release from non-degradable polymer matrices 48 2.4.3.1. Effects of monomer and polymer type and properties on drug diffusion through and drug release from non-degradable polymer matrices 48 2.4.3.2. Effects of drug loading on drug release from nondegradable polymer matrices 50 2.4.3.3. Effects of y-irradiation on drug release from non-degradable polymer matrices 52 2.5. Selection of polymers for perivascular films 53 2.5.1. Rational for selection of polymer for perivascular films 53 2.5.2. Poly(ethylene-co-vinyl acetate) (EVA) 54 2.5.3. Polyurethane 57 . EXPERIMENTAL 60 3.1. Materials 60 3.1.1. Drugs 60 3.1.2. Polymers 60 3.1.3. Nomenclature of polymers 60 3.1.4. Washing and storage of polymers 61 3.1.5. Manufacture of films 61 3.1.6. Phosphate buffered saline with albumin 61 3.1.7. Gel permeation chromatography 62 3.1.8. HPLC mobile phase 62 3.2. Equipment 62 3.2.1. Balances 62 3.2.2. Washing and storage of polymers 63 3.2.3. Manufacture of films 663 3.2.4. Differential scanning calorimetry 63 3.2.5. Thermogravimetric analysis 64 3.2.6. X-ray diffraction 64 3.2.7. Determination of molecular weight 65 3.2.7.1. Viscometry 65 3.2.7.2. Gel permeation chromatography 65 3.2.8. Rheological analysis 65 3.2.9. Film thickness 65 3.2.10. Phosphate buffered saline with albumin 66 3.2.11. Diffusion studies 66 3.2.12. Drug release studies 66 3.2.13. Paclitaxel Extraction 67 3.2.14. Paclitaxel HPLC analysis 67 3.2.15. Statistical analysis 67 3.3. Methods 68 3.3.1. Washing and storage of polymers 68 3.3.2. Preparation of films 69 3.3.3. Differential scanning calorimetry (DSC) 70 3.3.4. Thermogravimetric analysis (TGA) 70 3.3.5. X-ray diffraction (XRD) 70 3.3.6. Determination of molecular weight 71 3.3.6.1. Viscometry 71 3.3.6.2. Gel permeation chromatography (GPC) 71 3.3.7. Determination of rheological properties 72 3.3.8. Thickness of Films 72 3.3.9. Visual estimation of the solubility of paclitaxel in EVA and polyurethane... 72 3.3.10. Phosphate buffered saline with albumin 73 3.3.11. Characterization of paclitaxel diffusion in EVA and polyurethane films 73 3.3.12. Characterization of paclitaxel release from EVA and polyurethane films 74 3.3.13. Validation of paclitaxel calibration curves used for HPLC analysis 75 3.3.14. Paclitaxel analysis 76 3.3.15. Statistical analysis 76 . RESULTS 79 4.1. The physicochemical properties of paclitaxel loaded EVA and polyurethane films 79 4.1.1. Thermal analysis of films 79 4.1.2. X-ray 84 4.1.3. Molecular weight determination of polymers 85 4.1.3.1. Viscometry of films 85 4.1.3.2. Gel permeation chromatography of films 87 4.1.4. Rheology of films 91 4.1.5. Thickness of films for drug release studies 94 4.1.6. Visual estimation of the solubility of paclitaxel in EVA and polyurethane... 95 4.1.7. Paclitaxel calibration curves 96 4.2. The effect of polymer type on the diffusion of paclitaxel through EVA and polyurethane films 99 4.3. The effect of monomer and polymer type on the release of paclitaxel from EVA and polyurethane films 101 4.4. The effect of drug loading on the release of paclitaxel from EVA and polyurethane films 102 4.5. Modeling of drug release from paclitaxel loaded EVA and polyurethane films 104 5. DISCUSSION 108 5.1. The physicochemical properties of paclitaxel loaded EVA and polyurethane films 108 5.2. Experimental variables in paclitaxel diffusion and release studies with EVA and polyurethane films 114 5.3. The effect of polymer type on the diffusion of paclitaxel through EVA and polyurethane films 118 5.4. The effect of monomer and polymer type on the release of paclitaxel from EVA and polyurethane films 120 5.5. The effect of drug loading on the release of paclitaxel from EVA and polyurethane films 124 5.6. Selection of polymer for controlled release of paclitaxel for perivascular application for the inhibition of restenosis 125 6. SUMMARY AND CONCLUSIONS 127 6.1. Physicochemical characterization of paclitaxel loaded 60/40 EVA, 72/28 EVA, and polyurethane films 127 6.2. Paclitaxel in vitro diffusion through 60/40 EVA and polyurethane films 128 6.3. Paclitaxel in vitro release from 60/40 EVA, 72/28 EVA, and polyurethane films 128 viii 6.4. Selection of polymer for paclitaxel loaded films for perivascular application for the inhibition of restenosis 129 7. FUTURE WORK 130 appendix i: sample calculations of diffusion parameters 131 8. REFERENCES 135 ix. LIST OF TABLES Table 1. Melting temperature peak (Tm) of 5, 10, and 30 % w/w paclitaxel loaded 60/40 EVA and 72/28 EVA films, nonsterile and sterilized. Values (°C) are the mean ± standard deviation of the measurements of five samples. .82 Table 2. Degree of crystallinity of 5, 10, and 30 % w/w paclitaxel loaded 60/40 EVA and 72/28 EVA films, nonsterile and sterilized. Values (%) are the mean ± standard deviation of the measurements of five samples 82 Table 3. Intrinsic viscosity data. Standard polystyrene values represented are a mean calculated from three measurements ± standard deviation 88 Table 4. The M G P C of 60/40 EVA, 72/28 EVA, and polyurethane from nonsterile films 89 Table 5. Characteristic complex relaxation time (A*) for paclitaxel loaded 60/40 EVA, 72/28 EVA, and polyurethane films 83 Table 6. The thickness (h) of 5, 10, and 30 % w/w paclitaxel loaded 60/40 EVA, 72/28 EVA, and polyurethane films, nonsterile and sterilized. Values are the mean ± standard deviation of the measurement of five samples 95 Table 7. Solubility of paclitaxel in 60/40 EVA and polyurethane expressed as % w/v as determined by various methods 96 Table 8. Accuracy of paclitaxel calibration curves. These curves were determined on day one of four days. Data for days two, three, and four are not shown 98 Table 9. Representative intraassay precision and summary of interassay precision of paclitaxel calibration curves 99 Table 10. Coefficient values obtained from the cumulative diffusion of paclitaxel through 60/40 EVA and polyurethane films 100 Table 11. Diffusion coefficient values, D^and D, of paclitaxel in 60/40 EVA, 72/28 EVA, and polyurethane films 124 X LIST OF FIGURES Figure 1. A perivascular film, wrapped around the adventitia of an artery and secured in place with a single suture 5 Figure 2. A schematic of a cross-section of A) a normal artery and B) a restenotic artery 7 Figure 3. The chemical structure of paclitaxel. Adapted from (Wani, 1971) 13 Figure 4. Wave-forms for oscillatory strain {y0), rate of strain (y0*), and stress (<70) for an elastic body. Both strain and stress are in phase and lie on the solid line, while rate of strain is completely out of phase and is represented with the dashed line. Adapted from (Barnes, 1989) 31 Figure 5. A schematic of a plot of drug diffusion in a diffusion cell through a polymer film. As according to Equation 41, M i s plotted as a function of t, the slope of the straight line at steady-state diffusion is equal to PSCd, and tL is taken from the extrapolation of the straight line to the x-axis 43 Figure 6. A schematic of a non-degradable polymer matrix and its receding boundary as drug diffuses from the dosage form. As the drug is released from the matrix then the thickness of the matrix (h) through which the drug diffuses increases with time as the boundary between the matrix with drug and without drug recedes from the surface of the matrix. The total concentration of undissolved and dissolved drug is given by (A) as compared to the solubility (saturation concentration) of the drug as given by (G"). The receding boundary moves left by an infinitesimal distance (dh) as the infinitesimal amount of drug (dQ) is released. (Adapted from (Martin, 1993)) ......45 Figure 7. The chemical structure of poly(ethylene-co-vinyl acetate) (EVA). E V A is almost a perfectly random copolymer consisting of ethylene and vinyl acetate monomers 54 Figure 8. The chemical structure of SG85A polyurethane, poly(poly(tetramethylene ether glycol)-co-methylene bis(cyclohexyl) diisocyanate)-co-1,4-butanediol). This structure can be divided into soft and hard segments. The soft segment is composed of poly(tetramethylene ether glycol) (PTMEG), while the hard segment consists of methylene bis(cyclohexyl) diisocyanate (HMDI) and the chain extender, 1,4-butanediol 58 Figure 9. DSC thermograms of films prepared from A) polyurethane, B) 60/40 E V A , and C) 72/28 E V A , heated at 10 °C-min"1 81 xi Figure 10. TGA thermograms of films prepared from A) 72/28 EVA, B) 60/40 EVA, and C) polyurethane, heated at 60 °C-min_1 83 Figure 11. XRD patterns of films prepared from A) 60/40 EVA, B) 72/28 EVA, and polyurethane, scanned at 5° 29-min"1 at increments of 0.02° 20 84 Figure 12. Huggins (Equation 4) and Kraemer (Equation 5) plots of viscosity data for A) polyurethane, B) 60/40 EVA, and C) 72/28 EVA in chloroform at 25 ± 0.5 °C. Y-intercept values of each line are the respective Huggins ([r\]u) or Kraemer ([r|]K) intrinsic viscosity of the polymer 86 Figure 13. GPC universal calibration curve of polystyrene molecular weight (MW) standards 89 Figure 14. GPC elution profiles of a) nonsterile and b) sterilized films prepared from A) 60/40 EVA, B) 72/28 EVA, and C) polyurethane 90 Figure 15. Representative rheology plots for nonsterile films prepared from A) 72/28 EVA, B) 60/40 EVA, and C) polyurethane, and for sterilized, 30 % paclitaxel loaded films prepared from D) 72/28 EVA, E) 60/40 EVA, and F) polyurethane. The films were maintained at 80 °C during analysis 93 Figure 16. In vitro cumulative diffusion of paclitaxel through A) 60/40 EVA and B) polyurethane films. The points on the graph are the mean of three samples, and the error bars represent one standard deviation. The lines are the linear regression analysis lines of the four terminal points, extrapolated to the x-axis 100 Figure 17. In vitro cumulative release of paclitaxel (in u,g) from films prepared from A) 60/40 EVA, B) 72/28 EVA, and C) polyurethane. The films were loaded with a) 5, b) 10, and c) 30 % w/w paclitaxel. The points on the graph are the mean of five samples, and the error bars represent one standard deviation 103 Figure 18. In vitro cumulative release of paclitaxel (in \xg) as a function of the square root of time, from films prepared from A) 60/40 EVA, B) 72/28 EVA, and C) polyurethane. The films were loaded with a) 5, b) 10, and c) 30 % w/w paclitaxel. The points on the graph are the mean of five samples, and the error bars represent one standard deviation. The lines on the graphs are the linear regression analysis lines of the points 105 Xl l Figure 19. In vitro cumulative release of paclitaxel (in (ig) at 5 days as a function of the square root of loading concentration from films prepared from A) 60/40 EVA, B) 72/28 EVA, and C) polyurethane. The films were loaded with 5, 10, and 30 % w/w paclitaxel. The points on the graph are the mean of five samples, and the error bars represent one standard deviation. The lines on the graphs are the linear regression analysis lines of the points 107 Figure 20. Schematic of proposed complexation of paclitaxel with EVA and polyurethane. The dashed lines represent hydrogen bonds between the drug and the polymers 123 LIST OF ABBREVIATIONS A total concentration of undissolved and dissolved drug loaded in a matrix Ao amount of drug in buffer initially Ax, amount of drug in buffer at equilibrium A C N acetonitrile b a constant BSA bovine serum albumin c a constant C concentration Cd concentration of drug in a donor compartment Cpo concentration of drug initially loaded in polymer matrix Cr concentration of drug in a receiving compartment Cy solubility of drug in buffer or in matrix Cs • solubility of drug in polymer determined from diffusion study Cj- solubility of drug in polymer determined from diffusion and release studies Cj •• • solubility drug in polymer determined from visual study C V coefficient of variation cx concentration of polymer in chloroform Ci concentration of drug inside a membrane at the edge of the membrane beside the donor compartment C2 concentration of drug inside a membrane at the edge of the membrane beside the receiving compartment d density D diffusion coefficient DCM dichloromethane De Deborah number Deff effective diffusion coefficient DSC differential scanning calorimetry E ethylene E Young' s modulus EEA poly(ethylene-co-ethacrylate) EVA poly(ethylene-co-vinyl acetate) F force G Gibbs free energy G' storage modulus G" loss modulus G* complex shear modulus GPC gel permeation chromatography GTP guanidine triphosphate xV h haSMC HDPE HEMA HPLC i LP. I.V. J k K LDPE Lo L, m M MG?c height or thickness or distance human arterial smooth muscle cell high density polyethylene poly(hydroxyethyl methacrylate) high pressure liquid chromatography degree of polymerization or chain length intraperitoneal intravenous flux rate constant partition coefficient low density polyethylene original length final length slope cumulative amount of drug released from a matrix or diffused through a membrane gel permeation chromatography molecular weight number average molecular weight M, amount of drug in the donor compartment x v i MS MW mass spectometry number average molecular weight NMR P P PBSA PE PTCA PTFE PVA PVP Q RCAM S S.C. SEM SG85A SMC molecular weight Avrami exponent nuclear magnetic resonance permeability coefficient applied pressure phosphate buffer solution with albumin polyethylene percutaneous transluminal coronary angioplasty polytetrafluoroethylene poly(vinyl alcohol) Poly(vinyl pyrrolidone) cumulative amount of diffusant released from a matrix per surface area rat carotid artery model surface area or entropy subcutaneous scanning electron microscopy solvent grade, 85A shore hardness smooth muscle cell xvii t time tL lag time T temperature Tg glass transition temperature T°g symptomatic value of the glass transition temperature at infinite chain length TGA thermogravimetric analysis tL theoretical lag time Tm melting temperature Tr retention time tx time for solution of polymer in chloroform to flow through viscometer to time for chloroform to flow through viscometer U internal energy v volume VB volume of buffer Vd donor compartment volume PV volume at constant pressure VA vinyl acetate VSMC vascular smooth muscle cell WE membrane weight xvii i WL mass of the melt at time t wo mass of the melt at time zero x distance or thickness XRD X-ray diffraction 8 phase shift or phase lag of strain relative to stress Sv change in volume AG change in free energy AHf enthalpy of fusion AHfioo% enthalpy of fusion of theoretically 100% crystalline polymer AL length of extension AS change in entropy e porosity s extensional strain £ rate of extensional strain Y shear strain / rate of shear strain Yo strain amplitude Yo' rate of strain amplitude ?] viscosity [ rj\H Huggins intrinsic viscosity [ 7]]MK mean of the Huggins and Kraemer intrinsic viscosities for a given sample [ r/]K Kraemer intrinsic viscosity [ Tj]sc Solomon and Ciuta intrinsic viscosity A* characteristic time-dependent relaxation time cr shear stress GE in phase stress or "elastic stress" <7v out of phase stress or "viscous stress" <Jo stress amplitude r relaxation time co frequency co* characteristic oscillation frequency ACKNOWLEDGEMENTS I thank my supervisor, Dr. Helen Burt, for her guidance over the course of my M.Sc. studies. I thank my committee members for their direction of my research project: Drs. Jack Diamond, Colin Fyfe, Bob Miller, Kishor Wasan, and a special thanks to Sawaas Hatzikiriakos for the use of the rheological equipment. Thank you to Dr. Eugene Rosenbaum for technical assistance with the rheological equipment. I thank my colleagues in the lab for their helpful discussions of my research project: John Jackson, Dr. Richard Liggins, Chuck Winternitz, Ruiwen Shi, Jingfang Wang, and Dechi Guan. Thank you to Kevin Letchford for help with the rheological studies and to Allyson Bousquet for help with the manufacturing of films. I thank Dr. Bill Hunter, Dave Hartnett, Dr. Pierre Signore, Myra Lankford, and Leslie Crookall, of Angiotech Pharmaceuticals, Inc., for their insightful comments on my research project, and for their help with coordinating the sterilization of the films. Thank you to Angiotech Pharmaceuticals, Inc. for their financial support. I thank Orville Wright, Wilbur Wright, Richard Bach, and Dave Parry, for their inspiration and providing me with a means to pursue my other passion. Thank you to Charlotte Springate (Mom) and Gordon Springate (Dad), for their support of my M.Sc. studies. 1 1. INTRODUCTION A major complication of percutaneous transluminal coronary angioplasty (PTCA) and coronary artery bypass grafting is restenosis of the treated coronary artery (Peng, 1996; Mishaly, 1997). This intimal hyperplasia is due to proliferation of connective tissue and vascular smooth muscle cells at the site of vascular injury (Lambert, 1994; Mishaly, 1997). The process involves medial smooth muscle cells changing from a contractile to a migrative, proliferative form. Microtubules are required for this differentiation (Sollot, 1995). The new phenotype migrates to the arterial intima from the media and subsequently synthesizes and excretes an extracellular collagenous matrix , forming a neointimal layer. Enclosure of the arterial lumen by the neointima may then compromise the coronary flow (Mishaly, 1997). Continuous, local administration of an agent that inhibits restenosis at doses too low to provide an effect i f administered systemically, provides the benefit of local anti-restenotic treatment without systemic side effects. When administered from the perivascular site (the outer adventitial layer of the blood vessel) to the vessel wall at the site of balloon injury, these benefits may be achieved (Edelman, 1996). A schematic of a perivascular film wrapped around the adventitia of an artery is shown in Figure 1. Over the last several years, perivascular drug delivery systems have delivered drugs that reduce stenosis in arteries in animal models. Drug delivery to the adventitia of the blood vessel may target events occurring in the adventitia, media, and intima (Huehns, 1996). Poly(ethylene-co-vinyl acetate) (EVA) matrices loaded with drug have been applied perivascularly to effectively deliver a number of different drugs locally with reduced 2 systemic side effects, including heparin (Edelman, 1990), antisense oligonucleotides (Edelman, 1995), acidic fibroblast growth factor (Sellke, 1996), a tyrphostin (Golomb, 1996), colchicine (Mishaly, 1997), and paclitaxel (Sollot, 1995). Effective long-term anti-restenotic therapy may be required to prevent wound healing that occurs over weeks or months (Schwartz, 1996). Paclitaxel stabilizes microtubules and in this way inhibits differentiation o f vascular smooth muscle cells, from a contractile phenotype to a migratory and proliferative phenotype, in vitro. In vivo, paclitaxel has inhibited vascular smooth muscle cell migration and proliferation in rabbits when delivered by microporous balloons (Axel , 1997) and in the rat carotid artery model when administered intravenously (I.V.) (Sollot, 1995). Paclitaxel has been shown to inhibit vascular smooth muscle cell migration and proliferation in vitro for 14 d at a concentration of 0.1 to 10 umol-L"1, incubating over only a 20 min period (Axel , 1997). Paclitaxel loaded non-biodegradable polymeric implantable films are being investigated for perivascular application for local controlled delivery of paclitaxel in vivo to inhibit restenosis. Notwithstanding the potential of this novel drug delivery system, the mechanisms and release characteristics of paclitaxel from non-biodegradable matrices are not fully understood. The release of drug from polymer matrices is dependent on several factors. These factors include the physicochemical characteristics of the drug and the polymer, such as the crystallinity of the polymer, the molecular weight of the polymer, the solubility of the drug in the polymer, and whether pores or channels are created during the release of the drug. Another factor influencing the release o f drug 3 from polymer matrices is the amount of drug loaded into the matrices. Variation of polymer type or of monomer ratios produces different hydrophobic and crystalline properties of the polymer (Banberger, 1989). Therefore, paclitaxel matrix formulations with potentially different release characteristics may be developed from polymers with different compositions. Ideal features of a perivascular wrap for drug delivery to the adventitia include flexibility, controlled release of the drug, biocompatibility, and sterility. EVA and the following polyurethane, poly(poly(tetramethylene ether glycol)-co-mefhylene bis(cyclohexyl) diisocyanate-co-1,4 butane diol), were chosen for investigation in this work. The biocompatibility of EVA is established (Langer, 1981) as is the biocompatibility of polyurethane (Joshi, 1996), which is a requirement for implantation in vivo. EVA and polyurethane are hydrophobic polymers and therefore their matrices are expected to release a hydrophobic drug like paclitaxel into aqueous media in a slow, controlled manner. Both of these polymers are non-biodegradable, which may allow for the controlled release of a drug over several months or years. EVA and polyurethane are elastic polymers, permitting the manufacture of flexible matrices which may be wrapped around the vasculature. There are little data published in the literature regarding paclitaxel loaded EVA or polyurethane matrices for perivascular application for the inhibition of restenosis. In this project, EVA of two different monomer ratios, 60/40 E/VA and 72/28 E/VA, and polyurethane, were used to formulate matrices loaded with paclitaxel, to determine the effects of polymer composition on release rate characteristics. The effects of EVA 4 monomer ratio, polymer type, and drug loading on other properties such as thermal properties, the degree of polymer crystallinity, and mechanical properties were also investigated. In this work, it was hypothesized that the polymer type , monomer ratios of EVA, and drug loading in the film or wrap would influence the paclitaxel release profiles from the perivascular films. The purpose of this work was to develop a paclitaxel loaded drug delivery system for perivascular application to inhibit restenosis and to enhance our understanding of both the transport of paclitaxel through non-degradable polymer matrices and the release of paclitaxel from non-degradable polymer matrices. Thus, the specific aims of this project were to: 1 ) Characterize the effects of polymer type on the diffusion of paclitaxel through EVA and polyurethane films. 2) Characterize the effects of EVA monomer ratio and polymer type on the physicochemical properties and drug release profiles of paclitaxel loaded EVA and polyurethane films. 3) Characterize the effects of drug loading on the physicochemical properties and drug release profiles of paclitaxel loaded EVA and polyurethane films. 4) Characterize the effects of sterilization by y-irradiation on the physicochemical properties of paclitaxel loaded EVA and polyurethane films. 5) Select a polymer with which to manufacture paclitaxel loaded perivascular films for future evaluation in an animal model of restenosis. 5 Figure 1. A perivascular film, wrapped around the adventitia of an artery and secured in place with a single suture. 6 2. BACKGROUND 2.1. Restenosis 2.1.1. Anatomy of arterial wall A schematic of a cross-section of a normal artery is shown in Figure 2A. Blood vessel walls consist of three main layers: the innermost layer surrounding the lumen is the tunica intima, proceeding outward from the intima is the tunica media, and the outermost layer of the vessel wall is the tunica adventitia. In large arteries, the intima is made up of endothelial smooth muscle cells (SMCs) which are surrounded by a matrix of collagen, proteoglycans, and smaller amounts of elastin. In small vessels and capillaries, the intima consists of a single layer of endothelial cells and is referred to as the endothelium. The media is made up of multiple layers of SMCs, and in elastic arteries, these SMCs are surrounded by elastin. The adventitia is organized to permit vasoconstriction and dilation. It consists of collagen and elastic fibers, SMCs, fibroblasts, small blood vessels that produce the vasa vasorum, and nerves (Schwartz, 1996; Huehns, 1996). adventitia 8 2.1.2. Mechanisms of restenosis A schematic of a restenotic artery is shown in Figure 2B. Histological restenosis may be defined as neointimal hyperplasia at the site of percutaneous transluminal coronary angioplasty (PTCA); angiographic restenosis as the development of significant luminal narrowing at the treatment site; and clinical restenosis as the late recurrence of signs or symptoms of ischemia after an initially successful angioplasty. The term, restenosis, may refer to one or a combination of these definitions (Kuntz, 1993). Geometric remodeling may be defined as the change in the total vessel circumference in response to angioplasty or during atherogenesis (Post, 1995). The process of vascular injury following angioplasty can be separated into three phases (Faxon, 1997). The thrombotic phase is first, which involves platelet adherence and aggregation, and the initiation of the inflammatory response. The granulation phase is second, which involves SMC migration, proliferation, and secretion of an extracellular matrix. The remodeling phase is last, which involves reendothelialization and matrix organization. Clowes et al. showed that in the rat carotid artery model, which involves balloon denudation of the external carotid artery, luminal narrowing at 2 weeks after injury was in large part due to smooth muscle contraction of the vessel, and at 4 to 12 weeks due only to intimal thickening (Clowes, 1983a). SMC proliferation reached a maximum in the media at 2 d and in the intima at 4 d. SMC proliferation had decreased to normal levels at 4 weeks in the media and by 8 weeks in the intima that was covered by endothelium. SMC proliferation continued at a high level at the surface of the intima lacking endothelium at 12 weeks (Clowes, 1983b). During the period between 2 to 12 weeks, the fraction of the intima occupied by smooth muscle cells decreased markedly, but the total volume due to SMCs remained relatively constant (Clowes, 1983a). These data support the concept that intimal SMC proliferation after arterial injury is an acute event related to the initial injury process. Persistent proliferation of intimal SMCs did not result in an increase in intimal cell number (Clowes, 1983b). The results indicated that continued intimal thickening at late time points was due to synthesis and accumulation of connective tissue without further increase in SMC number (Clowes, 1983a). It has been shown that remodeling at the reference site may lead to an underestimation of the luminal narrowing in restenotic lesions when reference site diameters are used to determine luminal narrowing (Kakuta, 1998). Heller (1995) showed that matrix metalloproteases played a role in extracellular matrix remodelling of the neointima in the rat carotid artery model (RCAM) following balloon injury (Heller, 1995). Bennett (1994) showed that c-myc expression is required for vascular SMC proliferation in vitro and in the vessel wall of the rat carotid artery (Bennett, 1994). Edelman (1995A) confirmed Bennett's work with c-myc and showed that c-myb expression is required for VSMC proliferation in vivo in the RCAM with balloon injury (Edelman, 1995). Golomb (1996). showed that tyrosine phosphorylation was enhanced in the RCAM with balloon injury (Golomb, 1996). Edelman (1992) showed that bFGF could be released continuously from basic fibroblast growth factor (bFGF) ionically bound to heparin-Sepharose beads encapsulated 10 within calcium alginate microcapsules. Increased cell proliferation (not extracellular matrix production) and the intimal hyperplastic response in the RCAM with balloon injury (no effect was observed in the absence of vessel injury) were observed. It was found that bFGF induced angiogenesis within, and surrounding, the release device and in the vasa vasorum around the injured arteries. The effect of bFGF on SMC proliferation was related to its effect on the vasa vasorum. Thus, it was found that the in vivo angiogenic and mitogenic properties of bFGF are related and may be affected by local injury, or by factors in the vessel wall (Edelman, 1992). In vitro, it was shown that macrophage-colony-stimulating factor (M-CSF), M-CSF receptor protein, and M-CSF receptor gene transcripts, play a role in vascular SMC proliferation. M-CSF also modulated the expression of c-fos, c-myc, egr-1, and junB, all proto-oncogenes. M-CSF also acted together with thrombin, PDGF, or bFGF in promoting VSMC deoxyribonucleic acid (DNA) synthesis (Herembert, 1997). Barker (1995). showed that removal of the adventitia from the carotid artery of normal rabbits resulted in intimal lesions at 14 d, consisting of SMCs, and by 28 d the lesions had regressed. In similar rabbits that had been fed a high-cholesterol diet, at 14 d, the intimal lesions consisted of a mixture of macrophages and SMCs, and by 42 d, the macrophages were predominant (as foam cells). In rabbits continuously fed a high-cholesterol diet, the intimal lesions did not regress. This model showed that there are different responses to vascular injury in normal or hypercholesterolemic animals (Barker, 1995). 11 "It is evident that microtubules are involved in the control of the most critical and sensitive intracellular mechanisms necessary for vascular SMCs to undergo the multiple transformations involved in the development of atherosclerosis and restenosis after arterial injury, making microtubules particularly strategic targets to influence the outcome" (Kinsella, 1995). Agents that influence SMC motility may positively influence remodeling (Faxon, 1997). 2.1.3. Pharmacological interventions and drug delivery systems for restenosis Heparin and non-anticoagulating heparin, released from EVA matrices, inhibited SMC proliferation in the RCAM at 14 d after balloon injury (Edelman, 1990). In the same model, it was found that the administration of bFGF did not alter the ability of heparin to inhibit SMC proliferation (Edelman, 1992). AG-17, a tyrphostin, is one of a series of low molecular weight protein tyrosine kinase inhibitors. When released from EVA matrices, AG-17 was shown to inhibit neointimal formation in the RCAM with balloon injury, as measured by intima/media areas ratio (Golomb, 1996). Galardin, a peptide analogue, has been shown to inhibit neointimal formation as measured by intima/media areas ratio in the RCAM (Heller, 1995). In the RCAM with balloon denudation, the topical application of antisense oligodeoxynucleotides (AS-ODNs) to cdc2 and cdk2, resulted in reductions of the intimal/medial cross-sectional area ratio of 47 and 55%, respectively (Abe, 1994). The in vitro administration of AS-ODNs to c-myc, which expresses a proto-oncogene, to cultures of vascular SMCs reduced average cell levels of c-myc mRNA and protein by 50 to 55% and inhibited vascular SMCs from proliferating. The AS-ODNs to c-myc had an 12 IC5o of 10 lag-mL"1. Using the RCAM, after balloon injury, c-myc mRNA expression peaked at 2 h. AS-ODNs to c-myc, applied perivascularly, reduced peak c-myc expression by 75% and inhibited neointimal formation at 14 d (Bennett, 1994). The in vitro administration of AS-ODNs to c-myc and c-myb reduced SMC growth by approximately 55 %. When the AS-ODNs were applied perivascularly in the RCAM at the time of balloon injury, the AS-ODNs to c-myb inhibited intimal hyperplasia 2 weeks after injury, whether released from short time course F127 gels (i.e. released over several days) or from longer time course EVA matrices (i.e. released over several weeks). However, the AS-ODNs to c-myc inhibited intimal hyperplasia 2 weeks after injury, only when released from the longer time course EVA matrices and not when released from the short time course F127 gels (Edelman, 1995). That is, the duration of release of the drug influenced its effect on the inhibition of the intimal hyperplasia. AS-ODNs to the rat tenascin mRNA were shown in vitro to inhibit SMC proliferation and migration (Cleek, 1997). Kinsella et al. demonstrated in vitro, that nanomolar levels of paclitaxel inhibited PDGF-mediated vascular SMC migration. This inhibition occurred primarily via interference of cell locomotion and/or shape changes. The authors also demonstrated that paclitaxel inhibited in vitro vascular SMC division and platelet-derived growth factor-BB (PDGF-BB) stimulated c-fos messenger riboxynucleic acid expression. That is, inhibition of immediate early gene induction was another important mechanism by which paclitaxel blocked growth factor stimulation in vascular SMCs. The authors concluded that the abnormal microtubule polymerizing and stabilizing action of paclitaxel was responsible for the functional changes observed in the vascular SMCs (Kinsella, 1995). 13 2.2. Paclitaxel The structure of paclitaxel is shown in Figure 3. Figure 3. The chemical structure of paclitaxel. Adapted from (Wani, 1971). o 2.2.1.1. Rationale for selection of paclitaxel as an inhibitor of restenosis In a rabbit model of restenosis, intimal plaques were produced by electrical stimulation, and the animals then underwent balloon injury. In this model, paclitaxel administered by microporous balloons, reduced the intimal wall area, wall thickness, and the degree of restenosis compared with controls (Axel, 1997). In a balloon injured rat carotid artery model, treated with paclitaxel in a DMSO:Cremophor EL:ethanol vehicle injected intraperitoneally, the neointimal area was reduced by 70% (Kinsella, 1995). In vivo, paclitaxel inhibited vascular SMC migration and proliferation in the rat carotid artery model (Sollot, 1995). 14 2.2.1.2. Mechanism of action of paclitaxel as an anti-restenotic agent Paclitaxel inhibits the G2-M phase in the cell cycle and blocks mitosis. Other naturally occurring microtubule spindle poisons, such as vincristine, vinblastine, colchicine, podophyllotoxin, work by binding to soluble tubulin and inhibiting the polymerization of tubulin to form microtubules (Rose, 1995). While paclitaxel is a mitotic inhibitor, it is unique in that it promotes the assembly of tubulin into calcium-stable microtubules in vitro in the presence or absence of GTP or microtubule-associated proteins (Schiff, 1979), and stabilizes tubulin polymers against depolymerization back to tubulin (Schiff, 1980). This is the opposite effect of the above agents. Paclitaxel thus disrupts microtubule bundles and causes abnormal mitotic spindle aster production and microtubule bundling (Rose, 1995). Due to paclitaxel's effects on microtubules, not only does the drug appear to block mitotic spindle formation during cell division, it also disrupts the integrity of interphase microtubules that are necessary in other cell functions (Rowinsky, 1992). In this way, paclitaxel inhibits differentiation of vascular SMCs from a contractile phenotype to a migratory and proliferative phenotype in vitro (Sollot, 1995). Incubations of paclitaxel with human arterial SMCs in vitro, resulted in complete and prolonged inhibition of the cells' growth. When mitogens such as platelet-derived growth factor-AB (PDGF-AB), basic fibroblast growth factor (bFGF), or thrombin, were added to the human arterial SMC culture, they did not attenuate the effects of paclitaxel. The microtubule network of the human arterial SMCs was shown by immunohistochemistry to have undergone characteristic changes. That is, paclitaxel inhibits human arterial SMC proliferation and migration by interfering with the 15 microtubule network (Axel, 1997). It has been shown that microtubule stabilization is also critical in paclitaxel inhibition of vascular SMC migration in vitro (Kinsella, 1995). Loss of the acetyl group at C-l 0 (10-deacetyltaxol) or replacement of the benzyl group at C-5' with C(CH3)=CHCH3 (Cephalomannine) results in microtubule activity being retained. Replacing the ester side chain at position C-l3 with various functional groups, including CH 3 (Baccatin III), CH 2OH (19-hydroxybaccatin III), OH (O-cinnamoyltaxicin-I triacetate), and H (O-cinnamoyltaxicin-II triacetate), results in loss of all activity, suggesting that the ester side chain is required for activity. The loss of the acetyl group at C-l 0 had no effect on the activity of paclitaxel on microtubule assembly. Acetylation at C-2' and C-7 (2',7-diacetyltaxol) results in the loss of microtubule assembly activity in vitro, but these compounds retain their activity in vivo, possibly because they are converted intracellularly to paclitaxel or paclitaxel-like compounds (Parness, 1982). 2.2.1.3. Toxicities of paclitaxel When administered intravenously, the dose limiting side effect of paclitaxel is usually neutropenia (Rowinsky, 1995). Hypersensitivity reactions, mucositis (Rowinsky, 1995), neurotoxicity (Kelsen, 1994; Rowinsky, 1995), and myelosuppression (Kelsen, 1994) have also presented as dose limiting side effects of paclitaxel. Other toxic effects of paclitaxel include anemia, granulocytopenia (Ajani, 1995), alopecia, diarrhea, nausea, vomiting, thrombocytopenia, myalgia and cardiac toxicity (Rowinsky, 1995). 16 2.2.1.4. Pharmacokinetics of paclitaxel 2 Six and twenty-four hour infusions of paclitaxel (dose of 275 mg-m ) resulted in peak plasma concentrations capable of inducing biological and cytotoxic effects in vitro (i.e. 0.1 - 10 umolL '). Reported mean values of volumes of distribution ranged from 49 -2 to 119 L-iri at steady-state. These large volumes of distribution for paclitaxel indicate that there is extensive binding of the drug to circulating proteins and/or a tissue -1 -2 component. Paclitaxel was cleared from plasma in the range of 8 - 53 L-h -m . The half-life of elimination (t1/2) of paclitaxel ranged from 1.4 - 8.0 h (Rowinsky, 1992). High paclitaxel levels and hydroxylated metabolites have been observed in both the rat and human bile (Monsarrat, 1990; Monsarrat, 1992). Rowinsky et al. showed that less than ten percent of intraveneously administered paclitaxel was recovered in the urine after 1 - 2 d. In vitro incubations of human arterial SMCs with single doses of paclitaxel at nanomolar levels for 20 min, 24 h, or 14 d, resulted in complete and prolonged inhibition of cell growth for up to 2 weeks. The IC50 was determined to be 2.0 nmol-L"1 (Axel, 1997). Concentration and duration of exposure both effected the induction of in vitro microtubule and cytotoxic effects. The high molecular weight, bulky structure, and hepatic metabolism of paclitaxel suggest that after intraperitoneal administration, drug exposure in the peritoneal cavity may be greater than systemic exposure. Intraperitoneal delivery of paclitaxel at doses of 25 - 200 mg-m_2 resulted in peritoneal fluid levels of drug approximately one thousand 17 fold higher than those of plasma levels. After only a single intraperitoneal dose of paclitaxel, concentrations persisted in the peritoneal cavity for at least 48 h (Markman, 1992). 2.2.2. Physicochemical properties of paclitaxel Wani et al. first reported the structure of paclitaxel, then called taxol, in 1971. Paclitaxel was isolated from the stem bark of the western yew, Taxus brevifolia. Paclitaxel was the first agent with a taxane ring that was observed to have antitumor and antileukemic properties (Wani, 1971). Paclitaxel has a molecular weight of 853.9 g-mof1 (Rizzo, 1990) and is a modified diterpenoid with several acetyl groups and an additional complex ester side chain at the C-l3 position. Diterpenoids are derived biogenetically from geranylgeraniol pyrophosphate and have a twenty carbon skeleton (Kingston, 1991). Paclitaxel is a hydrophobic compound with both anhydrous and dihydrate forms (Dordunoo, 1996). The dihydrate has a water solubility of 1 (ig'mL"1 at 37 °C (Liggins, 1997). Paclitaxel undergoes hydrolysis in water with pseudo first-order kinetics. Maximum stability at 37 °C in buffer was observed at pH 3 to pH 5 (Dordunoo, 1996). Paclitaxel and its 1-epi isomer, 7-ep/'-taxol, exist in equilibrium with each other in solution (McLaughlin, 1981). 18 2.3. Polymers 2.3.1. Morphology of polymers The properties of a polymer are dependent on its three dimensional structure. The constitution, configuration, and conformation of the polymer can define the morphology of a polymer. Constitution refers to the types of atoms and functional groups which make up a polymer. Configuration describes how the atoms of a polymer are arranged about the polymer backbone atoms. Conformation relates how the configurations within polymer chains produce the three dimensional arrangement of the polymer in space. The polymerization process of the polymer generally determines constitution and configuration. Although conformation is associated with constitution and configuration, it is dependent on the environment of the polymer. That is, a polymer in an aqueous medium may have a different conformation than if it were in a hydrophobic medium. 2.3.1.1. Polymer constitution Polymer constitution describes the atoms in the chain, the type of end groups and side groups, the monomers' sequence, the type and size of branch units, and the molecular weight and the distribution of molecular weight. Most synthetic polymers have carbon atoms in their backbone chain, with the exception of the polysilicones. Other atoms that may be included in a carbon backbone chain are nitrogen and oxygen. The end groups of polymers may consist of the end groups of the original monomers, such as acids and amides in Nylon® synthesized by step-growth reactions. They may consist of radicals or ions in the growing chains of addition reactions; or they may consist 19 of a stable endgroup as the result of the termination of a growing addition reaction chain (Grulke, 1994). Homopolymers are polymers that contain only one type of repeating unit. Even condensation polymers made from two different monomer types, such as a dicarboxylic acid and a diamine, form only one type of repeating unit, and therefore are homopolymers. Copolymers are polymers that contain two or more different repeating units. If the different types of repeating units are randomly placed, then the polymer is referred to as a random or statistical copolymer. If linear chains are formed that contain long blocks from two or more different monomer types, but each block contains only one monomer type, then the polymer is referred to as a block copolymer. If branches of polymer chains are grafted onto linear chains that have a different type of repeating unit, then the polymer is referred to as a graft copolymer (Grulke, 1994). Chain branching results in a less ordered polymer chain that packs less densely than linear polymers. This is why polyethylene (PE) polymers with a greater number of branches are referred to as low density polyethylene (LDPE) and those with a lesser number of branches are called high density polyethylene (HDPE). Besides decreasing the density of a sample, increasing chain branching also affects the melt viscosity. This is because branched chains start shear thinning at lower shear rates compared to linear chains. However, stiffening of the chains due to chain branches produces higher limiting viscosities (Moore, 1984). 20 2.3.1.2. Polymer configuration Configuration depicts in detail how the bond angles and ease of rotations between atoms in a polymer chain leads to the orientation of atoms or groups about atoms within the chain. The sp3 orbitals of carbon form a tetrahedron with angles of 109° 29 . Carbon-carbon bonds along polymer backbones have bond angles around these tetrahedral angles, ranging from 105° to 113° (Hall, 1981). Monomers of the H2C=CHX type usually polymerize in a manner that results in the X group being attached to every second carbon atom in the backbone. There are three different ways that the X groups can be arranged in relation to the backbone plane. Each of these three ways that the X groups can be arranged in relation to the backbone plane is a type of stereoisomer. The isotactic stereoisomer exists when the polymer chain is in the all trans conformation and each X group is lined upon the same side of the backbone plane. The syndiotactic chain occurs when the X groups alternate on either side of the plane. The atactic stereoisomer is the result of a random arrangement of the X groups. Each of these stereoisomers is an extreme case. That is, a polymer is rarely purely isotactic, syndiotactic or even 100% randomly arranged in the atactic form. Most polymers are a mixture of two or more of these stereoisomers (Cowie, 1973). 2.3.1.3. Polymer conformation Conformation describes the three dimensional arrangement of a polymer, which is dependent on the configurations within the polymer and on the environment in which the polymer exists. With no external ordering forces acting on a single chain, its most 21 probable conformation is actually a random coil (Hall, 1981). Other conformations of single chains include the helix and folded chain (Grulke, 1994). Block copolymers can achieve a conformation quite different from that of linear homopolymers. Thermoplastic elastomer block copolymers may be arranged in a conformation such that they have areas of aggregation of hard, glassy blocks, and areas of soft, rubbery blocks (Cowie, 1973). 2.3.2. Polymer molecular weight 2.3.2.1. Definitions of polymer molecular weights Molecular weight (MW) is dependent on the type of monomers(s) and method of synthesis of the polymer. The MW of different polymers can range from as low as approximately 300 g-mol"1 to as large as 1028 g-mol"1. MW distributions can vary from narrow to broad, and from Gaussian to asymmetric (Hall, 1981). As the MW of a polymer is increased, its density, melting temperature (Hall, 1981) and viscosity are also increased (Moore, 1984). Increasing MW and therefore the length of a polymer, increases the number of secondary bonding forces between chains. Thus, it requires more energy (i.e. higher temperature or increased shear) to cleave the increased number of secondary bonding forces that exist with increased MW (Moore, 1984). 22 2.3.2.2. Measurement of polymer molecular weight As a polymer sample is likely to consist of individual polymer molecules with a distribution of molecular weights, several parameters are used to describe the distribution. The number average molecular weight (Mn) is given by ^ Equation 1 where «, is the number of molecules with the number of repeat units / and Mt is the molecular weight of the polymer chains with degree of polymerization /. The weight average molecular weight (Mw) is given by M Equation 2 where w, is the weight fraction of the molecules with the number of repeat units /. The distribution can be represented by the polydispersity index (PDI) which is given by M PDI = — - Equation 3 M. 2.3.2.2.1. Viscometry of polymers A viscometer can be utilized to measure the viscosity (rf) of a polymer sample in a suitable solvent at a constant temperature. Huggins and Kraemer plots can then be constructed for each polymer using solutions for each polymer in a range of dilute 23 concentrations from zero to approximately 1.0 percent w/v. The plots are then made according to the following equations (Huggins, 1942) VsP/c = [ti] + k'[n]2c Equation 4 and (Kraemer, 1938) ln(/ft /c) = [n]+k'[rj]2c Equation 5 where [rj] is the intrinsic viscosity, ^ s p is the specific viscosity, rjr is the relative viscosity, and c is the concentration of the polymer solution. As the concentration of the polymer solution approaches zero, the values of r/sp I c and ln(/7r / c) approach [n]. Thus the Huggins intrinsic viscosity ([?7]H)and the Kraemer intrinsic viscosity ([TJ]K) are then taken as the intercept value of the linear regressions for each set of plots (that is, c = 0). The value reported as [/7]HK is the mean of the determined values of [TJ]H and [TJ]K-The [ rf] of a polymer sample can also be determined by utilizing the Solomon and Ciuta (Solomon, 1962) equation and the value for this method is symbolized by [rj]sc 2.3.2.2.2. Gel permeation chromatography (GPC) A universal calibration curve can be determined by plotting the log of the product of the molecular weight and [rj] of each of the standards against their retention times. The MQPC of the polymer samples can be calculated from the universal calibration curve using the equation [77] s c ~ (2 1 / 2 / cx) ((tx I to) - (lnfe /10)) 1/2 Equation 6 24 M G P C = 10{mTr'b) I [ 77] Equation 7 where m is the slope and b is the intercept of the universal calibration curve, and Tr is the retention time of the sample. 2.3.3. Polymer crystallinity and crystallization Crystalline regions of a polymer are formed by the solidification of portions of chains into regularly repeating internal arrangements. Conversely, amorphous polymers are formed by the solidification of the chains into a disordered, random internal arrangement. 2.3.3.1. Polymer crystallinity Polymers rarely form perfect crystals. Instead, solid polymers consist of small regions of three-dimensional order, called crystallites, which are interspersed among amorphous regions. Polymers that are not completely amorphous nor completely crystalline, but rather have some regions of crystallinity, are referred to as semicrystalline polymers. The density of the crystallites is uniform throughout a polymer but is different from the density of the amorphous regions. The greater the relative amount of crystalline regions to amorphous regions, the greater the crystallinity of a polymer. The degree of crystallinity of a polymer is defined as the percent weight of crystalline regions per weight of polymer sample. The general mechanical response, modulus, density, and optical clarity of a polymer are dependent upon its crystallinity (Tobolsky, 1971). 25 An ordered, regular polymer chain structure is required to pack the chains into a crystal lattice. Accordingly, stereoregular polymers are generally more crystalline than polymers with irregular structures (Tobolsky, 1971). Thermal energy has a disordering effect on polymers. Thus, the secondary bonding forces must have a high enough energy of interaction to hold the chains together in the crystal lattice against thermal disordering. Polymers with a stereoregular arrangement and many hydrogen bonds or van der Waals forces therefore tend to have a raised crystalline melting point (Cowie, 1973). Polymers with an irregular structure tend not to pack well into a crystal lattice and are generally amorphous. For example, the amorphous LDPE packs poorly (and thus its low density) because it has relatively more side chains, compared to the more crystalline HDPE with fewer side chains (Grulke, 1994). Completely amorphous polymers are transparent, while semicrystalline polymers are usually translucent or opaque. This is because some light is scattered when it passes between two phases that have different refractive indices (if the dimension of the discontinuities are the same or greater than the wavelength of visible light). Light is scattered continuously throughout a semi-crystalline polymer because of the difference in the refractive indices of the amorphous regions and the crystallites. Completely amorphous polymers on the other hand are a homogeneous material and have a constant refractive index throughout (Moore, 1984). 26 2.3.3.2. Measurement of polymer crystallinity Because crystallinity affects the melting temperature, density, modulus, and optical properties of polymer samples, these all form methods of determining the degree of crystallinity of a sample. By determining density or modulus as a function of temperature, the relative crystallinity of a series of semicrystalline samples of the same polymer type can be determined. Samples with a greater modulus or density, possess a greater degree of crystallinity. Semicrystalline polymers of the same type with higher melting points, will be more crystalline (Cowie, 1973). Optical and scanning electron microscopy are qualitative methods of determining whether there is crystallinity in a sample. The crystallinity of samples can also be determined by infrared (IR) spectroscopy or NMR (Hall, 1981). In general, perfect crystals result in very narrow X-ray diffraction peaks. The more amorphous material in a sample, the more the peaks are broadened. A measure of relative crystallinity for other samples of the same type of polymer can be obtained by comparing the intensities of the diffraction peaks (Hall, 1981). Polymer melting is observed as an endothermic event. The degree of crystallinity (reported as percent crystallinity of theoretically 100% crystalline polymer) of a polymer can be calculated by measuring the enthalpy of fusion (AHJ) and dividing this value by the value of AHf of theoretically 100% crystalline polymer (AHnbo%) according to the equation (Runt, 1985) %crystallinity = ( AH, ^ ^AH f m % j bcl00% Equation 8 27 2.3.4. Glass transition temperature (7^ ) of polymers Amorphous polymers exhibit one of two types of mechanical behavior . They may act as a glass, that is, they appear to be hard and rigid, but still flow under moderate stress. Alternatively, they may act rubbery, that is, they are flexible and soft. When glasses are heated, they eventually reach a temperature where they become rubbery. Similarly, when rubbery polymers are cooled they eventually reach a temperature where they become glasses. The temperature, or rather the relatively narrow range of temperatures above which a polymer is rubbery and below which it is a glass, is called the glass transition temperature (Tg) (Grulke, 1994). 2.3.4.1. Factors affecting 7^  of polymers When heating a glassy polymer, the observed Tg increases proportionally with an increase in the heating rate. The difference in Tg with different heating rates is due to the polymer chains being unable to respond instantaneously to the change in temperature. The higher the rate of heating (or cooling) the farther from equilibrium the observed Tg will be. Similarly, the amount of time that a glass is allowed to equilibrate after being rapidly cooled below Tg affects the observed Tg upon reheating the glass. The less time that a rapidly cooled glass is allowed to reach equilibrium, the higher the observed Tg will be upon reheating (Tobolsky, 1971; Cowie, 1973). 28 Introducing crosslinks into a polymer increases its density and decreases its molecular motion. As the number of crosslinks in a polymer is increased the Tg is increased and the transition zone around it is broadened. As the chain length or the molar mass of a given type of polymer is increased, so is its Tg. The value for Tg is given by Tg = T°g - CU Equation 9 where T°g is the value of the glass transition temperature at infinite chain length, C is a constant for the particular polymer, and / is the degree of polymerization or chain length. For most commercial polymers, the decrease in Tg with decrease in i is negligible due to the high degree of polymerization and thus Tg « T°g (Cowie, 1973). Solvents act on polymers in ways somewhat similar to heat. Solvents that can form strong secondary bonding forces with polymer chains, can replace the polymer interchain secondary bonding forces. The result is that as the solvent penetrates the polymer, the chains are pulled apart, thereby dissolving linear and branched polymers. Thus, as the concentration of solvent is increased, the Tg is decreased. When a blend is produced from two miscible polymers, the Tg of the blend will lie in between the Tg of each of the separate types. When two immiscible polymers are blended, the resulting blend will have two glass transitions (Grulke, 1994). 29 2.3.5. Mechanical properties of polymers 2.3.5.1. Types of mechanical deformations of polymers There are three basic types of mechanical deformations that can be applied to polymer solids. These deformations are simple extension (tensile), simple shear, and bulk compression. 2.3.5.1.1. Simple extension (tensile) deformation of polymers Simple extension or tensile occurs when stress (cr) is applied normally to the surface of the material. Extensional strain (s) is equal to the length that the sample has been extended (AL) divided by the original length of the sample (Lo) (Hosford, 1993). e = AL/Lo = (Li - Lo) /Lo Equation 10 where Lj is the length of the sample at time (t). Rate of extensional strain (£•*) is given by s = (1/Lo) (dLi/dt) Equation 11 Tensile strength is given by the highest value of stress on a stress-strain curve. Ductility is a measure of elongation or reduction in cross-sectional area. Materials that are more ductile will form a neck, while less ductile materials fracture before forming a neck (Hosford, 1993). 30 2.3.5.1.2. Simple shear deformation of polymers Simple shear occurs when stress is applied tangentially to the surface of the material (Cogswell, 1981). In simple shear, there is a change in shape and no change in volume (Sharma, 1965). Shear strain (y) is equal to the change in distance (JC) divided by the height (h). y = x/h Equation 12 Rate of shear strain (y) is given by / = (1/h) / (dy/dt) Equation 13 By oscillating the simple shear stress one can test for stress and strain response over a range of frequencies {co). The typical sinusoidal variations of stress amplitude (oo), strain amplitude (yo) and rate of strain amplitude (y0*) response for an elastic body in a dynamic experiment is given in Figure 4. The strain response of a perfect elastic material is completely in phase with the stress, while the rate of strain is completely out of phase (Sharma, 1965; Cogswell, 1981). 31 Figure 4. Wave-forms for oscillatory strain (y0), rate of strain (y0*), and stress (ao) for an elastic body. Both strain and stress are in phase and lie on the solid line, while rate of strain is completely out of phase and is represented with the dashed line. Adapted from (Barnes, 1989). 2.3.5.1.3. Bulk compression deformation of polymers The third type of mechanical deformation, bulk compression, occurs when stress is applied normal to all faces (Cogswell, 1981). In bulk compression, there is a change in volume and no change in shape (Sharma, 1965). The stress is the applied pressure (P) and the strain is the change in volume per unit volume (dv/v) (Cogswell, 1981). 2.3.5.2. Viscoelasticity of polymers Polymers usually show the behaviour of both elastic solids and viscous liquids. That is, due to the intermixing of elastic and viscous effects polymers show a retarded elastic deformation or viscoelastic state (Sharma, 1965). 32 2.3.5.2.1. Definition of elastic body A perfect elastic solid, also called a Hookean solid, is one in which stress (a) is directly proportional to extensional strain (s) or shear strain (y) but is independent of the rate of strain (Sharma, 1965). acce ox crocy Equation 14 Stress may be plotted against strain and Young's modulus (E) is equal to the slope of the stress-strain curve or stress divided by strain E - o~/y Equation 15 where stress is equal to the applied force (F) per unit area (A). o~ = F/A Equation 16 When stress is applied to an elastic body it results in instantaneous deformation. The elastic deformation is due to the bonds between atoms being stretched by a relatively small stress. When the stress is removed, the material returns to its original shape. Once the deformation state is reached there is no further deformation as long as the stress remains (Barnes, 1989; Cogswell, 1981). An elastic body can be modelled with a Hookean spring (Turner, 1993). At low deformations, the behaviour of rubbers approach that of Hookean, while many polymer melts have a lower modulus than that of rubbers (Sharma, 1965). 33 2.3.5.2.2. Definition of viscous liquid A perfect viscous liquid, also called a Newtonian liquid, is one in which stress is directly proportional to rate of strain and not to strain itself (Barnes, 1989; Cogswell, 1981). ax/ Equation 17 That is, deformation will continue as long as the stress is applied. This can be modelled with a Newtonian dashpot (Turner, 1993). By oscillating a polymer melt under simple shear stress, one can test for stress and strain responses over a range of frequencies. The rate of strain response of a perfect viscous material is completely in phase with the stress, while the strain is completely out of phase. 2.3.5.2.3. Definition of viscoelastic fluid A material is viscoelastic when stress is related to both strain and rate of strain (Hosford, 1993). A viscoelastic material can be modelled with the Maxwell model, which consists of a Hookean spring and a Newtonian dashpot linked in series. 2.3.5.2.4. Regions of viscoelasticity of polymers Plotting the elastic modulus of a polymer against an increase in temperature shows, in order of increasing temperature, a glassy state, a leathery state, a rubbery state, a rubbery flow, and a viscous flow. The transitions from state to state are a result of the molecular motion within the polymer. In the viscous melt or rubbery state, the chains are 34 in relatively rapid motion. As the temperature is decreased below the glass transition temperature (Tg) there is not enough potential energy to allow for most molecular movements and the chains are frozen in the conformation they were in when Tg was reached. At temperatures below Tgi the structure of an amorphous polymer is random and the resultant glassy state is equivalent to a frozen liquid. In the glassy state, the polymer acts like an elastic solid under stress, due to almost no molecular motion except the vibrations of atoms about equilibrium positions. As the temperature is increased, there is an increase in the motion of side chain groups and small groups of approximately four or five atoms along the main chain. In the leathery state, the molecular motions increase with increase in temperature while the modulus drops significantly. Above Tg there is cooperative jumping and wriggling of entire molecules up to 40 to 50 carbon atoms long. This allows for uncoiling and flexing which in turn result in elasticity. In the rubbery state, the modulus decreases only slightly with an increase in temperature. In the rubbery flow state, there is cooperative motion resulting in elasticity. This is followed by translational motion of entire molecules and thus under strain the initial elasticity is followed by flow. The modulus decreases as the temperature is increased. In the viscous state, there is constant translational motion of entire molecules which results in the polymer behaving as a viscous liquid with little elastic recovery. The modulus steadily decreases as the temperature is increased (Cowie, 1973). 35 2.3.5.3. Dynamic mechanical analysis (DMA) of polymers The advantage of dynamic sinusoidal shear and measurement of stress response is that it gives very precise information about polymer melts compared to tensile testing. Besides giving precise data over a wide frequency range, dynamic studies of polymer melts give direct measurements on both elastic and viscous properties, even at low viscosities (Cogswell, 1981). The mechanical behaviour of a polymer melt may be examined under dynamic testing between two oscillating plates or cones. Since no polymer melt is perfectly viscous nor perfectly elastic, neither rate of strain nor strain are completely in phase with stress. Rather, the response of a polymer melt involves a phase shift or phase lag of strain relative to stress (S). A viscous response will often be seen at low co and an elastic response will often be seen at high co (Sharma, 1965; Cogswell, 1981). The complex shear modulus (G*) of a polymer melt is equal to the total of the storage modulus, also known as the dynamic rigidity (G), and the loss modulus (G'). The storage modulus is due to elastic deformation and the loss modulus is due to viscous flow deformation (Barnes, 1989). G* = G'+G" Equation 18 Strain is related to frequency by y = yo COS, cot Equation 19 36 and strain rate by / - - yoOi SINcot Equation 20 The relaxation time (r) and viscosity (77) of the polymer melt can be introduced to define stress as a= (( rjcoyo)/(l + coW)) (COTCOScot- SINcot) Equation 21 By making SINcot equal to zero one obtains the "elastic stress" (<JE), that is, the part of the stress that is in phase with strain. This stress is given by GE = G'y Equation 22 and thus the storage modulus is given by G' = ( TJTCO2 ) / (1 + co2 r2) Equation 23 Similarly, by making COScot equal to zero one obtains the "viscous stress" (cry), that is, the part of the stress that is completely out of phase with strain. This stress is given by cry = (G "/ co) y Equation 24 The loss modulus is thus given by G"= ( Tjco) / (1 + co2 x2) Equation 25 37 At low frequencies, there is a slow deformation of the polymer and the deformation is primarily viscous. The loss modulus is thus significantly higher than the storage modulus. As the frequency, is increased the storage modulus generally increases and under fast deformation at high frequencies the deformation is primarily elastic. The storage modulus is thus significantly higher than the loss modulus. The relationship between the phase shift and the storage and loss moduli is given by (Barnes, 1989) TAN 8=G'/G" Equation 26 Whether a polymer behaves as a viscous liquid or as an elastic solid depends on the time scale of the stress and strain and the time required for the material's time-dependent mechanisms to respond (Rosen, 1993). A characteristic time-dependent relaxation time (A*) can be defined by the equation A* = 1 / (co*)2 Equation 27 where co* is the characteristic oscillation frequency at which G'= G". A large A* indicates a slow time-dependent elastic response, while a small A* indicates a rapid response. The Deborah number (De) is the ratio of the A* to the time scale of the deformation (ts) De = A* Its Equation 28 At small Deborah numbers the response will appear viscous (i.e. De« 1) and at large Deborah numbers the response will appear elastic (i.e. De » 1) (Rosen, 1993). 38 2.4. Polymeric drug delivery Polymers for drug delivery may be separated into water-soluble, biodegradable, and nonbiodegradable materials. Water-soluble polymers are generally employed for delivery of drugs from several hours to several days. These polymers dissolve in the body after becoming hydrated, ionized, or protonated. Biodegradable polymers generally release drugs from several days to more than one year. These polymers undergo degradation, often involving hydrolysis, before dissolving. Nondegradable polymers are utilized for extended drug delivery in the order of months to years. These polymers are essentially inert and remain in the body indefinitely unless removed (Dunn, 1991). 2.4.1. Rationale for the use of a polymer drug delivery system for the controlled, local delivery of an inhibitor of restenosis In order for a patient to derive maximum benefit from a drug , the drug should be delivered to its specific target site at a rate and concentration that permit optimal therapeutic efficacy while reducing any undesirable side-effects to a minimum (Mills, 1987). Advantages of sustained release formulations include a reduction in the total dose of drug required, a reduction in the frequency of administration, and a decrease in side effects (Thompson, 1960). For the inhibition of restenosis, it is not currently established whether a local drug delivery system consisting of a nondegradable polymer or a biodegradable polymer may be more applicable. However, long-term therapy may be required to inhibit the process of restenosis, which occurs over weeks or months (Schwartz, 1996). 39 2.4.2. Drug release from non-porous, non-degradable polymer matrices Higuchi has derived a model for the release of drugs from solid matrices. Two mechanisms of drug release from granular matrices are suggested. In the first mechanism, incorporated solid drug in a uniform matrix is presumed to dissolve into the matrix, diffuse through the matrix, and out into the surrounding medium. In the second mechanism, the surrounding medium enters the matrix phase through pores and intergranular spaces. The drug is presumed to dissolve into the fluid medium, diffuse through and from the matrix through the pores filled with the fluid medium. The assumption of this second mechanism is that intragranular diffusion is insignificant. Drug associated with the surface of the matrix may be released more rapidly than drug residing within the matrix (Higuchi, 1963). 2.4.2.1. Drug diffusion through polymer matrices Fick's First Law describes the rate of diffusion through unit area at a steady state of flow. Flux (J) is the amount of material (M) that is transported through a cross-section of area (S) per unit of time (t) and is given by the equation (Flynn, 1974) J = dM/S'dt Equation 27 Flux is also proportional to the concentration gradient (dC/dx) across a membrane and this is represented by Fick's First Law (Flynn, 1974) J = -D(dC/dx) Equation 28 40 where D is the diffusion coefficient of the diffusant, C is its concentration and x is distance of its movement perpendicular to the surface of the barrier. The negative sign indicates that the diffusant is transported from the region of high concentration to low concentration (Flynn, 1974). By combining the above two equations, one obtains (Flynn, 1974) J = dM/S -dt = -D(dC/dx) Equation 29 By taking Cj and C2 to be the concentrations of the diffusant inside the membrane at the edge of the membrane beside the donor and receptor compartments, respectively, and h to be the distance across the membrane, then dC/dx can be approximated with (Cj -C2)/h such that Equation 29 can be written as (Flynn, 1974) dM/S -dt = D((C, - C2)/h) Equation 3 0 and rearranged to give dM/dt = DS((C, - C2)/h) Equation 31 Although Ci and C2 within the membrane are difficult to determine, the concentrations of the diffusant within the donor (Q) and receptor (Cr) compartments are easily determined. The partition coefficient (K) is given by (Martin, 1993) K = Ci/Cd = C2/Cr and thus Equation 31 can be represented as Equation 32 41 dM/dt = DSK((Cd - Cr)/h) Equation 33 If the receptor compartment approaches perfect sink conditions then Cr~0 and (Flynn, 1974) dM/dt = DSKCd/h Equation 34 A permeability coefficient (P) can be defined as (Martin, 1993) P = DK/h Equation 35 and K can be calculated from the equation (Martin, 1993) K = hP/D Equation 36 such that dM/dt = PSCd Equation 37 The partition coefficient can be calculated by adding a specific volume of buffer (VB) with the diffusant to a specific weight of membrane (WE) of known density (d) and mixing until the diffusant reaches an equilibrium between the buffer and the membrane. The amount of diffusant in the buffer is measured initially (A0) and at equilibrium (Ax). The partition coefficient is then given by (Maurin, 1992) K=((A0- Aoo) Ws/d) / (AJVB) Equation 38 42 The solubility (saturation concentration) (Cs) of the drug in the polymer can be calculated from Sometimes D, K and h can not be determined independently in order to calculate P. If Cd remains relatively constant throughout time, however, by measuring the amount of penetrant (M) in the receiving sink of the diffusion cell at various times (t), and measuring Q and S, then P can be calculated from the slope of the linear portion of the plot of M versus t according to equation 40 (Martin, 1993). A constant rate of diffusion into the receptor cell may not be observed until after some time. An example is shown in Figure 5. The early stage of the curve in Figure 3 is non-linear and thus represents a non-steady state release of diffusant; the later stage is linear and thus represents a steady state release of diffusant. The steady state portion of the curve may be back extrapolated to the time axis. The point at which this extrapolation intersects with the time axis is called the lag time (t£). The 4 is the theoretical time that is needed for a diffusant to establish a linear or uniform concentration gradient within the membrane. If there is a time lag, then the straight line of Figure 3 may be represented by the modification of Equation 40 and P can be calculated from the slope of (Flynn, 1974) Cy • — KCd Equation 39 M = PSCdt Equation 40 M = PSCd(t-tr) Equation 41 43 where tL is obtained by extrapolating the straight line to the x-axis. The ti is also given by (Flynn, 1974) tL = h2/6D Equation 42 such that D is given by D = h2/6tL Equation 43 Figure 5. A schematic of a plot of drug diffusion in a diffusion cell through a polymer film. A s according to Equation 41, M is plotted as a function of t, the slope o f the straight line at steady-state diffusion is equal to PSCj, and tL is taken from the extrapolation of the straight line to the x-axis. Adapted from (Martin, 1993). 0 5 10 15 20 Time (hours) 44 2.4.2.2. Drug release from non-porous non-degradable polymer matrices If the amount of diffusant (dM) that diffuses through a surface area (5) of the matrix is represented by dQ so that dQ = dM/S, then Fick's First Law can be adapted from Equation 23 to be given as (Higuchi, 1960) dM/S -dt = dQ/dt = DCS »/h Equation 44 where Cs- is the concentration of the diffusant in the matrix determined from drug release studies. This equation can be used to explain the release of drug from a polymer matrix, where dQ/dt is the rate of drug released per unit surface area of the matrix. A two-dimensional schematic of drug release from a polymer matrix is shown in Figure 6. 45 Figure 6. A schematic of a non-degradable polymer matrix and its receding boundary as drug diffuses from the dosage form. As the drug is released from the matrix then the thickness of the matrix (h) through which the drug diffuses increases with time as the boundary between the matrix with drug and without drug recedes from the surface of the matrix. The total concentration of undissolved and dissolved drug is given by (A) as compared to the solubility (saturation concentration) of the drug as given by The receding boundary moves left by an infinitesimal distance (dh) as the infinitesimal amount of drug (dQ) is released. (Adapted from (Martin, 1993)). Static diffusion layer Surrounding aqueous layer Perfect sink h ~ dh As the drug is released from the matrix then the thickness of the matrix (dh) through which the drug diffuses increases with time as the boundary between the matrix with drug and without drug recedes from the surface of the matrix. The total concentration of undissolved and dissolved drug is given by (A) as compared to the solubility (saturation concentration) of the drug as given by (Cs). The infinitesimal amount of drug (dQ) released as the receding boundary moves left by an infinitesimal distance (dh) is given by (Higuchi, 1960) dQ = (A- (1/2)CS-)dh = ((2A - Cs••) / 2) dh Equation 45 46 From this equation the dQ can be substituted in equation 44 to give (Higuchi, 1960) DCS"/h = ((2A - CS") /2) dh /dt Equation 46 and rearranged to give dt = ((2A -CS")/ (2DCS -))hdh Equation 47 This can be integrated to give (Higuchi, 1960) t = ((2A -CS") /(2DCS-)) 2h2 + constant Equation 48 The integration constant can be evaluated when t = 0 and therefore h = 0, which gives (Higuchi, 1960) t = ((2A - Cs ••) h2) / 4DCS - Equation 49 and h = ((4dcs -t) / (2A - Cs-))m Equation 50 Equation 45 can be integrated to give (Higuchi, 1960) Q = ((2A -Cs-)/2)h Equation 51 By substituting h from Equation 50 into this equation one obtains Q = ((2A -Cs-)/ 2) ((4DCS -t) / (2A - C, -))1'2 Equation 52 and this can be rearranged to give the Higuchi equation (Higuchi, 1960) 47 Q = (D (2A - Q-) (Cs-t))m Equation 53 By differentiating this equation with respect to t we obtain the rate of release of drug (dQ/dt) at a given time (t) (Higuchi, 1960) dQ/dt = (1/2) ((D (2A - Cs») Cr) /1)m Equation 54 and if one makes A » Cs then (2A - Cs-) ~2A and Equation 54 reduces to (Higuchi, 1960) dQ/dt = (ADCS- / 2t)m Equation 55 Thus the rate of release of drug at a given time (dQ/dt) is proportional to the square roots of the following factors: the total amount of drug in the matrix, the solubility of the drug in the matrix, the diffusion coefficient of the drug in the matrix, and the inverse of time. Similarly, Equation 53 reduces to (Higuchi, 1960) Q = (2ADQ -t)1/2 Equation 5 6 Thus the total amount of drug released is proportional to the square roots of the following factors: the total amount of drug in the matrix, the solubility of the drug in the matrix, the diffusion coefficient of the drug in the matrix, and time. After Q is determined (and A and t are known) then Cs- can be obtained by rearranging Equation 56 to give Cs - = Q2 I 2ADt Equation 57 48 where D is determined from drug diffusion studies. Values of Cs- for each of the drug loadings may be determined by obtaining Q and t from the best fit curve of the cumulative drug release plotted against the square root of time. 2.4.3. Variables affecting drug diffusion through and drug release from non-degradable polymer matrices Variables affecting drug diffusion through, and drug release from, non-degradable polymer matrices include monomer and polymer type, drug loading, and y-irradiation of the matrices. 2.4.3.1. Effects of monomer and polymer type and properties on drug diffusion through and drug release from non-degradable polymer matrices The diffusion characteristics of monosubstituted benzoic acids, including benzoic acid, chlorobenzoic acid, methoxybenzoic acid, methylbenzoic acid, were studied in EVA polymers with differing monomer ratios. EVA monomer ratios included 0/100 ( p o l y v i n y l acetate)), 91/9, 82/18, 75/25, 67/33, and 60/40 E/VA. It was shown that the diffusion coefficients were in a narrow range, however, the permeability and partition coefficients increased nonlinearly with an increase in vinyl acetate content of the EVA (Maurin, 1992). The authors suggested that the diffusion process was limited by the partitioning of the various benzoic acids between the polymer and the buffer. The partitioning and diffusion of the benzoic acids in the polymer matrices were proposed to be dependent on hydrogen bond interactions between the vinyl acetate in the polymer and the carboxylic acids in the benzoic acids. It was also shown that the permeability 49 coefficients increased with decreasing ionization of the benzoic acids, which suggests that the non-ionized, neutral forms were responsible for diffusing across the EVA membrane. However, there was no significant difference in the cumulative amount of benzoic acids released with time by varying the pH, which suggested that optimizing the solubilities of the benzoic acids in the EVA by altering the pH in the EVA would not affect their diffusion. That is, the increase in solubility of the benzoic acids in the EVA was cancelled out by the decrease in the partition coefficient. The authors also found that using 2-propanol as the diffusion medium, resulted in an increase in the donor phase solubility of the acid, the diffusion rate, and the permeability (Maurin, 1992). It was also shown that increased VA content of EVA matrices resulted in increased permeation of nicardipine hydrochloride (Morimoto, 1988). Permeation of benzalkonium chloride and chlorhexidine, through films prepared from 60/40 EVA or silicone elastomer (Silastic®), has been investigated using a two-chamber diffusion cell. Release profiles of the two drugs from loaded solid polymer matrices in buffer were also characterized. The benzalkonium chloride was found to diffuse through and release from the two polymer matrices more rapidly than the chlorhexidine. The authors suggested that chlorhexidine permeated the matrices more slowly because of its low solubility in water (Saltzman, 1991). Polymer crystallinity may play a role in the release of drug from a polymer matrix. As the crystallinity of a polymer matrix is increased, the density of the matrix increases and the molecular mobility of the polymer chains and the drug decreases, and the matrix thus becomes less permeable to drugs (Pitt, 1990; Omelczuk, 1992). 50 MW may influence the diffusion and release of drug in a polymer matrix. The diffusion of liquid silicone molecules through silicone polymer was observed to decrease as the molecular weight of the polymer was increased (Chien, 1978). Poly(DL-lactic acid) of either 150 000 g-mol"1 or 450 000 g-mol"1 MW was formulated into microspheres loaded with sulphadiazine. The matrices prepared from the lower MW polymer released 80 % of the loaded dose after 90 days, while the matrices prepared from the higher MW polymer released only 40 % of the loaded dose in the same period (Wise, 1978). The MW of the poly(DL-lactic acid) may have influenced the release of sulphadiazine via diffusion through the polymer and also the release of the drug via degradation of the polymer matrix. As the molecular weight of a polymer matrix is increased, the matrix becomes less permeable to drugs, due to an increased density and a glass transition temperature. The increased density and glass transition temperature result in a lower molecular mobility of the polymer chains and drug (Omelczuk, 1992). 2.4.3.2. Effects of drug loading on drug release from nondegradable polymer matrices Langer and his co-workers (Langer, 1983) showed that increased loading of bovine serum albumin (BSA) in EVA not only increased the rate of drug release, but also increased the total fraction of drug released. They determined by microscopy that these macromolecules were released from EVA by diffusion via interconnected pores that were created as drug was released. The pore structure could be used to control the rate of release. Increased drug loading increased the porosity of the matrix. With a porosity of 51 less than approximately 0.3, not all of the drug was released, but with a porosity above 0.3, the release rates increased significantly and all of the drug was released (Siegel, 1989; Bawa, 1985). Quantitative image analysis showed a continuous polymer phase with different sized pores (Saltzman, 1987). A restricted pore geometry appeared to be responsible for the slower release rate than would be expected with a single size of pores (Siegel, 1990). The release of BSA and y-globulin from EVA appeared to occur via a percolation process and the total fraction of drug released was correlated to the porosity of the polymer matrix based on the percolation theory. The release rate and total fraction of drug released was lower for y-globulin than it was for BSA. The y-globulin had a lower buffer solubility than the BSA (Saltzman, 1989). Other drugs, such as dopamine (Freese, 1989) and progesterone (Vandelli, 1993), also show an increased rate of release from EVA with an increase in drug loading. The release of drug from a porous polymer matrix depends on several factors: the penetration of the surrounding liquid into the matrix, the solubility of the drug in the matrix and in the surrounding liquid, and the release of the drug through the pores. The porosity (e) of a matrix is the fraction that is made up of channels or pores into which the surrounding liquid can enter. The porosity represents the total porosity of the matrix, which is due to the initial porosity (so) that is present before drug is released, and the porosity that is a result of the drug being released and leaving behind channels. 52 2.4.3.3. Effects of y-irradiation on drug release from non-degradable polymer matrices Sterilization is a necessary requirement for an in vivo implant. There are several methods used to sterilize biomaterials. Ethylene oxide (EO) often causes toxicological problems due to residual EO. Moist or dry heat sterilization often causes hydrolysis and degradation of the biomaterial. y-irradiation may also cause changes in the properties of the material. In the presence of oxygen, y-irradiation produces free radicals in polymers, which are often rapidly converted into peroxidic radicals (for example, COO-) (Sintzel, 1997). Drugs may undergo oxidation and coupling to polymers when incorporated into matrices and irradiated (Volland, 1994). It has been suggested that polymers of the cross-linking type, at low irradiation doses, produce oxidative products on their chains and subsequent chain scission (Sen, 1995). y-irradiation in air may increase oxidative degradation and mechanical erosion with ageing (Besong, 1998). Other mechanical effects of y-irradiation related to chain scission include, a decrease in thermal stability, a decrease in toughness and elongation at break, and an increase in tensile strength at break. The increase in tensile strength at break is attributed to a decrease in crystallinity (Sen, 1995). y-Irradiation that causes chain scission (degradation) results in a decrease in Mw and may result in an increase or a decrease in drug release from drug loaded polymer matrices (Volland, 1994). yirradiation of polymers in gels or solutions has resulted in a decrease in viscosity (Adams, 1973; Martini, 1997) although ethanol protected against the decrease in viscosity, y-irradiation has also resulted in crosslinks and gel network 53 structures (Martini, 1997). It is felt that drug distribution and polymer properties influence the stability of the drug against y-irradiation (Volland, 1994). y-irradiation of EVA results in chain scission, crosslinking, and a decrease in crystallinity. There is a decrease in both elongation at break and toughness when y-irradiated, probably due to chain scission (Sen, 1995). y-irradiation of captopril and poly(D,L-lactide-co-glycolide) microspheres resulted in a reduction in the average molecular weight of the polymers. The release profile of the drug from the microspheres changed after y-irradiation. For some molecular weights, the release of drug increased and for others it decreased. There was no correlation between molecular weight and change in release rate of drug due to y-irradiation (Volland, 1994). 2.5. Selection of polymers for perivascular films 2.5.1. Rational for selection of polymer for perivascular films Preferred design features of a perivascular wrap for drug delivery to the adventitia, include an elastic or flexible wrap. The drug delivery system should provide for controlled drug release, ideally only to the blood vessel. The polymer employed in the wrap should be biocompatible and sterile, which are requirements for implantation in vivo. The hydrophobicity of paclitaxel is an important factor in the choice of polymer for use in a delivery device for the controlled release of paclitaxel. When paclitaxel, which is hydrophobic, is incorporated into a hydrophobic polymeric matrix, this would be 54 expected to produce a matrix that would potentially result in a slow, controlled diffusion of the drug into an aqueous environment. The nondegradable polymers, EVA and polyurethane, were chosen for investigation in this work. 2.5.2. Poly(ethylene-co-vinyl acetate) (EVA) The structure of poly(ethylene-co-vinyl acetate) (EVA) is shown in Figure 7. Figure 7. The chemical structure of poly(ethylene-co-vinyl acetate) (EVA). EVA is almost a perfectly random copolymer consisting of ethylene and vinyl acetate monomers. ethylene vinyl acetate H-[CH2CH2]n-random-[CH2CH(OOCCH3)]m-H The reactivity ratios of ethylene ( r i ) and vinyl acetate (r2) are 1.08 and 1.07, respectively. Thus, the reactivity ratio product (r\v2) of ethylene and vinyl acetate monomers is equal to 1.16. That is, r \ ~ X2 ~ r \?2 ~ 1. This means that both monomers favor each other almost equally and will be randomly copolymerized within the chain and will have a uniform chain to chain constitution over any degree of conversion. The result is a polyethylene with acetoxy groups randomly distributed (Zutty, 1964). 55 Above 50 percent w/w vinyl acetate content, the polymer is rubbery and below 5 % w/w vinyl acetate content, the polymer is only slightly different from low density polyethylenes. EVA has mostly butyl short-chain branches. Long-chain branches can exceed 50 carbons in length and are the result of intermolecular chain transfer activity. Increasing the vinyl aceate content decreases the crystallinity and there is an increase in the glass transition temperature (Tg). EVA is semicrystalline up to 40 % w/w vinyl aceate content. Because an ordered, regular polymer chain structure is required to pack a polymer's chains into a crystal lattice, the crystallinity of EVA is dependent on the relative amount of ethylene monomers present (Banberger, 1989). The melting point (Tm) of EVA therefore also decreases with an increase in VA content. The value of AH f ioo%PE is 293 J-g"1 (Runt, 1985). From 0 % to approximately 45 % w/w VA content, the density of EVA increases non-linearly. This is because, although vinyl acetate has a greater density than ethylene and the content of vinyl acetate is increasing, the crystallinity is simultaneously decreasing, which results in looser "packing" of the polymer chains. The decrease in crystallinity has a greater effect than the increase in density due to vinyl acetate and thus a plot of density versus vinyl acetate content produces a curved line. At greater than 45 % w/w vinyl acetate the polymer is completely amorphous and the density of the polymer increased linearly with the increase in vinyl acetate, due to compositional changes (Salyer, 1971). 56 EVA as received from the supplier, contains the antioxidant butylhydroxytoluene (BHT). Methods of extracting BHT from EVA have included washing in absolute ethanol. Total volume changes of ethanol were made every 24 h and the presence of BHT in the wash was detected using UV-VIS at a wavelength of 230 nm. The washes were continued until the absorbance was less than 0.03 absorbance units (Tamargo, 1990). EVA was selected as a polymer with which to develop perivascular films because of its controlled release properties and safety track record as a drug delivery implant. EVA is nondegradable, which may allow for the controlled release of a drug over several months or years. EVA is an elastic polymer, permitting the manufacture of flexible matrices, which may be wrapped around the vasculature. This polymer is sterilizable by ethylene oxide or y-irradiation and its biocompatibility is established (Langer, 1981). Using an in vivo rabbit cornea implant model, it has been shown that alcohol washed EVA and poly(hydroxyethyl mefhacrylate) (HEMA) were noninflammatory, while other polymers, such as polyacrylamide and poly(vinyl pyrrolidone) (PVP) produced significant inflammation, including edema, cellular infiltration, and neovascularization. Unwashed EVA and unwashed poly(vinyl alcohol) (PVA) produced mild inflammation. Washed PVA produced mild to no inflammation (Langer, 1976; Langer, 1981). 57 2.5.3. Polyurethane The structure of SG85A polyurethane, poly(poly(tetramethylene ether glycol)-co-methylene bis(cyclohexyl) diisocyanate)-co-l,4-butanediol), is shown in Figure 8. Typically, a biomedical polyurethane is synthesized by a two-step process. The first step consists of the polymerization of a polyol (for example, poly(tetramethylene ether glycol) (PTMEG)) and a diisocyanate (for example, methylene bis(cyclohexyl) diisocyanate (HMDI)). The second step is the extension of the PTMEG-co-HMDI chain by polymerization with a diol chain extender (for example, 1,4-butanediol). 58 Figure 8. The chemical structure of SG85 A polyurethane, poly(poly(tetramethylene ether glycol)-co-methylene bis(cyclohexyl) diisocyanate)-co-l,4-butanediol). This structure can be divided into soft and hard segments. The soft segment is composed of poly(tetramethylene ether glycol) (PTMEG), while the hard segment consists of methylene bis(cyclohexyl) diisocyanate (HMDI) and the chain extender, 1,4-butanediol. Although the 1,4-butanediol and PTMEG are made up of essentially the same unit, the PTMEG consists of a number (j) of repeating units. HMDI urethane urethane PTMEG HO- C - N H — ( O V -CH 2 - N H — C — O 4 C H 2 - C H 2 - C H 2 - C H 2 - O -O - C - N H — ( O v - C H 2 - ^ - N H - C — O — C H 2 - C H 2 - C H 2 - C H 2 - O O -H urethane urethane 1,4-butanediol HMDI 59 Polyurethane was selected as a polymer for the development of perivascular films for similar reasons to the selection of EVA, that is, polyurethane is hydrophobic, nondegradable, elastic, and sterilizable. The biocompatibility of polyurethane has been evaluated in vitro (Francois, 1996; Skarja, 1997; Bernacca, 1998; Mohsen, 1998) and in vivo (Joshi, 1996; Maurin, 1997). Increasing surface smoothness and hydrophilicity may reduce the risk of bacterial colonization and infection of implanted polyurethane (Francois, 1996). A selection of different polyurethane samples including Biomer®, Mitrathane®, PEU-2000, and PEU-100, implanted subdermally in rats for 60 d resulted in thin fibrous capsules surrounding the samples (Joshi, 1996). Mitrathane®, a hydrophilic polyurethane, was milled and sprinkled intraperitoneally. This resulted in acute inflammation for the first 7 d and was followed by degradation and phagocytosis by macrophages for the next 120 d. During the 180 d of the study, little chronic inflammatory response was observed in the surrounding tissues (Maurin, 1997). 60 3. EXPERIMENTAL 3.1. Materials 3.1.1. Drugs Paclitaxel, Hauser (Boulder, CO) 7-epi-taxol, Hauser (Boulder, CO) 3.1.2. Polymers Poly(ethylene-co-vinyl acetate), Polysciences (Warrington, PA) SG85A Polyurethane, Thermomedics (Woburn, MA) 3.1.3. Nomenclature of polymers Poly(ethylene-co-vinyl acetate) (EVA) is a random copolymer consisting of ethylene and vinyl acetate monomers. The relative amount of each monomer in a particular EVA polymer is expressed as a ratio. For example, EVA consisting of 60% w/w ethylene and 40 % w/w vinyl acetate is abbreviated as 60/40 EVA. SG85A is the manufacturer's abbreviation for solvent grade, 85A shore hardness polyurethane. 61 3.1.4. Washing and storage of polymers Chloroform, HPLC grade, Fisher Scientific (Fair Lawn, NJ) Ethanol (95%), Commercial Alcohols (Vancouver, BC) Silica gel desiccant, Fisher Scientific (Fair Lawn, NJ) 3.1.5. Manufacture of films Dichloromethane, HPLC grade, Fisher Scientific (Fair Lawn, NJ) Poly(ethylene-co-vinyl acetate), 60/40 and 72/28 % w/w, washed Polyurethane, SG85A, dried (refer to sections 3.1.2 and 3.3.1) Teflon® FEP film, Bytac type VF-81, Norton Performance Plastics (Akron, OH) Paper, econosource type, Unisource (Canada) Pouches, heat sealable 4.5 mils stock no. 500, Kapak® (Minneapolis, MN) 3.1.6. Phosphate buffered saline with albumin Albumin Fraktion V, Boehringer Mannheim (Germany) Sodium chloride, Fisher Scientific (Fair Lawn, NJ) Sodium phosphate dibasic heptahydrate, Fisher Scientific (Fair Lawn, NJ) Sodium phosphate monobasic, Fisher Scientific (Fair Lawn, NJ) 62 Water, distilled and deionized via a Milli-RO Water System, Millipore 3.1.7. Gel permeation chromatography Chloroform, HPLC grade, Fisher Scientific (Fair Lawn, NJ) Polystyrene standards, MW 30k and 50k, Polysciences (Warrington, PA) Polystyrene standards, MW 100k and 233k, Pressure Chemical Co. (Pittsburgh, PA) 3.1.8. HPLC mobile phase Acetonitrile, HPLC grade, Fisher Scientific (Fair Lawn, NJ) Water, distilled and deionized via a Milli-RO Water System, Millipore Methanol, HPLC grade, Fisher Scientific (Fair Lawn, NJ) Filters, type HV, 0.45 um, Millipore (Bedford, MA) 3.2. Equipment 3.2.1. Balances models AJ100, AE163, and PJ300, Mettler Corporation (Switzerland) 63 3.2.2. Washing and storage of polymers Beakers, 1 L, Pyrex brand Hot plate stirrer, model PC-351, Corning Vacuum oven, model 5831, Napco Vacuum pump, model SA55NXGTE-4870, Emerson (St. Louis, MO) 3.2.3. Manufacture of films Rotator, Labquake Shaker model, Barnstead/Thermolyne (Dubuque, Iowa) Pipettors, 5 mL Pipettman model, Gilson (France) Petri dish, Kimax® brand Vacuum oven, model 5831, Napco Vacuum pump, model SA55NXGTE-4870, Emerson (St. Louis, MO) Heat sealer, Impulse model AIE-300, American International Electric 3.2.4. Differential scanning calorimetry Lids, aluminum X1014, Rheometric Scientific Pans, aluminum XI021, Rheometric Scientific 64 Differential scanning calorimeter, model Pyris 1 DSC, Perkin Elmer (Norwald, CT) Liquid nitrogen cooling system, CryoFill model, Perkin Elmer (Norwald, CT) Computer, Venturis 575, Digital Software, Pyris Series, Perkin Elmer (Norwald, CT) 3.2.5. Thermogravimetric analysis Pans, aluminum XI021, Rheometric Scientific Thermogravimetric analyzer, model TGA 7, Perkin Elmer (Norwald, CT) Controller, TAC 7/DX Thermal Analysis model, Perkin Elmer (Norwald, CT) Computer, Venturis 575, Digital Software, Pyris Series Version 3.01 Revision A, Perkin Elmer (Norwald, CT) 3.2.6. X-ray diffraction X-ray diffractometer, Geigerflex model, Rigaku (Tokyo, Japan) 65 3.2.7. Determination of molecular weight 3.2.7.1. Viscometry Viscometer, 25 um, Canon-Fenske Circulating water bath, model B-1, MGW Lauda (Germany) 3.2.7.2. Gel permeation chromatography Vials, 1 mL, Hewlett Packard (Richmond, BC) Column, 104 A PLgel column, 5um gel beads, Hewlett Packard (Richmond, BC) Auto injector, model SIL-9A, Shimadzu (Tokyo, Japan) Refractive index detector, model RID-6A, Shimadzu (Tokyo, Japan) Chromatograph, model LC-10AD, Shimadzu (Tokyo, Japan) 3.2.8. Rheological analysis Rheometer, parallel plate, System IV model, Rheometrics (Piscatawa, JF) 3.2.9. Film thickness Micrometer, Mitutoyo model CD-6"B (Japan) 66 3.2.10. Phosphate buffered saline with albumin Erlenmeyer, 4 L, Kimax® brand Stirrer, Canlab® model H2165-20, American Hospital Supply Canada (Toronto, ON) 3.2.11. Diffusion studies Cells, donor and receiving, 20 mL borosilicate glass scintillation vials with urea lined screw caps, Kimble Glass (Vineland, NJ) Incubator shaker, Inova 4000 model, New Brunswick Scientific (Edison, NJ) 3.2.12. Drug release studies Hole punch, 6.2 mm diameter Micrometer, model CS-6", Mitutoyo (Japan) Culture test tubes, 15 mL with polytetrafluoroethylene (PTFE) lined screw caps, Kimax® Oven, Shel Lab model, Sheldon Manufacturing (Portland, OR) 67 3.2.13. Paclitaxel Extraction Evaporator, Reacti-vap III model, Pierce (Rockford, IL) Heating module, Reacti-therm III model, Pierce (Rockford, IL) Mixer, Maxi Mix II type 37600 model, Thermolyne (Dubuque, Iowa) 3.2.14. Paclitaxel HPLC analysis Vials, 1 mL HPLC, Waters (Milford, MA) Autosampler, 717plus model, Waters (Milford, MA) Controller, 600S model, Waters (Milford, MA) Detector, UV vis, 486 Tunable Absorbance Detector model, Waters (Milford, MA) Chromatograph, 746 Data Module model, Waters (Milford, MA) 3.2.15. Statistical analysis Software, SPSS 7.5 for Windows Student Version Hardware, Pentium II®, 300 MHz 68 3.3. Methods 3.3.1. Washing and storage of polymers BHT was removed from 60/40 EVA and 72/28 EVA by washing with DCM and ethanol. Twenty grams of each polymer were placed into separate one liter beakers. A six centimeter stir bar and 300 mL of DCM were added to each beaker. Each beaker was covered with a watch glass, placed on a hot plate stirrer, and stirred at low speed until all of the polymer was dissolved. The 60/40 EVA was dissolved at ambient temperature, while the 72/28 EVA was dissolved at approximately 40 °C to facilitate dissolution. After the polymer was dissolved, then 400 mL of ethanol was added (while continuing to stir). The polymer then precipitated out of solution as a gooey mass. The stirrer was turned off and the suspension left for ten minutes while most of the polymer settled to the bottom of the beaker. The solvent was then decanted and discarded. The polymer was then dissolved and precipitated three more times for a total of four washes. The precipitate was then transferred to a wash glass and dried in the fume hood for two days. The precipitate was then cut into approximately six millimeter cubes and placed under 20 in. Hg vacuum for one day. The washed polymers were then stored in a sealed container in the freezer until use. Polyurethane was dried by placing it under approximately 26 in. Hg vacuum for two days. The dried polymer was then stored in a sealed container with silica gel desiccant at ambient temperature until use. 69 3.3.2. Preparation of films Films for all studies (except the determination of paclitaxel solubility by visual observation) were prepared from 60/40 EVA, 72/28 EVA, or polyurethane and loaded with 5 to 30% w/w paclitaxel using a solvent casting technique. Polymer and paclitaxel having a total weight of two grams were added to a 20 mL scintillation vial and made up to 20 mL with dichloromethane (DCM) to make a ten percent weight per volume solution of polymer and paclitaxel in DCM. The vial was then capped and rotated end over end for approximately 12 h to allow the polymer and paclitaxel to dissolve. Vials containing 72/28 EVA were heated to 45 °C for this 12 h period to dissolve the polymer. To make films, two milliliters of solution were pipetted onto a 4 x 4 cm square of Teflon® film that had previously been adhered to a glass slab. The solution was pipetted in a fume hood with the fan off. The Teflon® square and solution cast on it were then covered with a Petri dish to decrease the evaporation rate of DCM from the film; the fume hood was turned on and DCM allowed to evaporate for six hours. The film was then dried further in a vacuum oven at ambient temperature under 26 in. Hg for approximately 12 h. Each film was then cut in half and each half was heat sealed in a 4 x 6 cm plastic pouch. One half of each film was then sterilized by E. H. & S. (California) by y-irradiation for 282 min at a dose rate of approximately 9 000 Rad-min"1, receiving approximately 2.5 MRad. The temperature of the material during irradiation was kept at or below four degrees Celsius by the use of blue ice cooling throughout the irradiation cycle. 70 Films for the transparency study were loaded with 0.1 to 30% w/w paclitaxel and manufactured using a similar solvent casting technique. These films were prepared by pipetting 250 uL of solution onto a one x two centimeter rectangle of Teflon®, to give the same volume of solution per area of Teflon® as in the preparation of the other films. 3.3.3. Differential scanning calorimetry (DSC) Pans (with lids crimped non-hermetically) holding approximately 5 mg of film, accurately weighed, were heated from -75 to 125 °C at 10 °C-min"'. The cell was purged with nitrogen gas flowing at 20 mL-min"1. Polymer melting was observed as an endothermic event. Reported values are presented as the mean ± one standard deviation of values observed for the analysis of four separate films. 3.3.4. Thermogravimetric analysis (TGA) Pans holding approximately 5 mg of film accurately weighed were heated at 60 °C-min"' from ambient temperature to 500 °C. 3.3.5. X-ray diffraction (XRD) The °29 range scanned was 5° to 120° at a rate of 5° 26-min"1 at increments of 0.02° 20. The X-ray source was CuKa radiation (40 kV, 20 mA). Films were loaded by laying them across the sample holder (no glass backing) and taping the edges of the film to the edges of the sample holder. 71 3.3.6. Determination of molecular weight 3.3.6.1. Viscometry The Solomon and Ciuta intrinsic viscosity ([r|]sc) of polystyrene molecular weight standards in the range of 30k to 233k g-mol"1 was determined for concentrations between 0.20 to 0.35 % w/v in chloroform using Equation 6. Reported values are presented as the mean + standard deviation of values observed for three replicate determinations of each polystyrene standard. Huggins and Kraemer plots (Equation 4 and Equation 5, respectively) were constructed for each polymer. The plots were constructed using four solutions for each polymer in a range of concentrations between 0.2 and 0.8 percent w/v in chloroform maintained at 25 ± 1 °C. Each solution was measured three times and the mean value plotted. The Huggins intrinsic viscosity ([^]H)and the Kraemer intrinsic viscosity ([^JK.) were taken as the intercept value of the linear regressions for each set of plots. The value reported as [/7]HK is the mean of the determined values of [TJ]H and [^ ]K-3.3.6.2. Gel permeation chromatography (GPC) Polystyrene standards were prepared as a 0.25% w/v solution in chloroform and assayed at ambient temperature with an injection volume of 10 uL, and a mobile phase of chloroform flowing at 1 mL-min"1. A GPC universal calibration curve was determined by plotting the log of the product of the molecular weight and [rj] of each of the standards against their retention times. 72 The GPC conditions for the polymer samples were the same as those for the standards. The GPC molecular weight (MGPC) of the polymer samples was calculated from the GPC universal calibration curve. Reported M G P C values are presented as the mean ± one standard deviation of values observed for five samples. 3.3.7. Determination of rheological properties The parallel plate rheometer was loaded with a 1 mm thick slab of polymer matrix with or without paclitaxel. Matrices 1 mm thick were constructed by "glueing" films together by placing a few drops of DCM between films and pressing them together by hand. The upper plate was oscillated from 10"1 to 103 rad-s"1 at 120 °C to calculate the storage modulus (G'), loss modulus (G") and complex viscosity (77*) of each sample. Reported values are presented as the mean ± one standard deviation of values observed for three samples. 3.3.8. Thickness of Films The thickness of each film was measured in different locations three times. Reported values are presented as the mean ± one standard deviation of values observed for the analysis of five separate films. 3.3.9. Visual estimation of the solubility of paclitaxel in EVA and polyurethane The solubility of paclitaxel in 60/40 EVA and polyurethane films was determined visually. A series of films with paclitaxel loadings ranging from 0.1 to 30 % w/w were examined. A transparent film was taken be a solution of paclitaxel in the polymer, while 73 an opaque film was taken to be a saturated solution of paclitaxel in the polymer. The two closest concentrations of drug loaded films in which the film with a lower concentration of drug was clear, and the film with a greater concentration of drug was opaque, were taken to bracket the solubility of paclitaxel in that polymer. The visually determined solubility of paclitaxel in the polymer (C") was reported as being between these two concentrations. 3.3.10. Phosphate buffered saline with albumin Phosphate buffered saline with albumin (PBSA) was prepared by adding 32.88 g sodium chloride, 8.60 g sodium phosphate dibasic heptahydrate, 1.26 g sodium phosphate monobasic, and 0.40 g albumin to a 4 L Erlenmeyer flask and making up to 4.0 L with water. The flask was stirred with a magnetic stir bar for one hour to allow the salts and albumin to go into solution, and then stored at four degrees Celsius until use. 3.3.11. Characterization of paclitaxel diffusion in EVA and polyurethane films In vitro cumulative diffusion of paclitaxel through 60/40 EVA and polyurethane films was plotted against time. The theoretical time lag (tL) was determined by extrapolating the linear regression analysis line of the four terminal points to the x-axis. The diffusion coefficient (£>) was calculated from the theoretical lag time with the film thickness (h) in Equation 43. The permeability coefficient (P) was calculated from the slopes of the linear regression analysis lines (Equation 41). The partition coefficient (K) was calculated from the h, P, and D in Equation 36. The solubility of paclitaxel in each 74 polymer (C5) was calculated from the K and the solubility of paclitaxel in the donor compartment (Q) in Equation 39. Diffusion studies were carried out using cells consisting of a donor cell and a receiving cell clamped on either side of a polymer film sample (no drug). The donor cell had 20 mL of a saturated solution of paclitaxel in PBSA at 37 °C and the receiving cell had 20 mL of PBSA at 37 °C. At various times the receiving cell was removed for paclitaxel analysis and to maintain sink conditions. The receiving cell was then replaced with another receiving cell containing fresh PBSA at 37 °C. Diffusion studies were carried out with 60/40 EVA and polyurethane films; however, the 72/28 EVA films were too fragile to complete the studies. Reported values of the cumulative amounts of paclitaxel diffused into the receiving cell are presented as the mean ± standard deviation of values observed for three samples. 3.3.12. Characterization of paclitaxel release from EVA and polyurethane films Sample films were cut out with a hole punch, accurately weighed and their thickness measured with a micrometer. The films were then tumbled end over end at 37 °C in 14 mL PBSA in culture tubes. At various time points the supernatant was removed for paclitaxel analysis and replaced with fresh PBSA at 37 °C to maintain sink conditions. Reported values of the cumulative amounts of paclitaxel released are presented as the mean ± standard deviation of values observed for five samples. 75 3.3.13. Validation of paclitaxel calibration curves used for HPLC analysis A calibration curve for paclitaxel was validated by measuring four sets of standards on four separate days. The linearity, accuracy, intraassay precision, and interassay precision of the calibration curve were measured. Linearity was expressed as the coefficient of determination (R2). Accuracy was expressed in terms of the average percent deviation, or bias, based on three replicate spiked paclitaxel solutions. The average deviation from the expected value of paclitaxel concentrations was measured from spiked samples for each of the nine concentration data points in the calibration curve. Mean values of each concentration within ± 15% of the actual value were taken as an acceptable accuracy (Shah, 1992). Intraassay precision was expressed in terms of the coefficient of variation (CV) of the mean of the four replicate data points on each day. Interassay precision was expressed in terms of the CV of the mean of the sixteen replicate data points collected over four days. A value of less than 15% is taken to be acceptable accuracy, except at the limit of quantitation, where it should not deviate by more than 20%. Values of precision around the mean value that were taken to be acceptable, were those not exceeding 15%, or for the limit of quantitation, those not exceeding 20% (Shah, 1992). Values of accuracy and precision are reported as the mean ± standard deviation. Calibration curves were prepared for paclitaxel dissolved in 60:40 ACN:water over a concentration range of 0.1 to 50 ng-mL-1. Paclitaxel was assayed at ambient temperature by HPLC with an injection volume of 20 uL and a mobile phase of acetonitrile:water:methanol 58:37:5 flowing at one milliliter per minute. The detection wavelength was 232 nm. 76 3.3.14. Paclitaxel analysis Paclitaxel was extracted by adding one milliliter DCM to the withdrawn PBSA supernatant and shaking for five seconds to allow the paclitaxel to partition into the organic phase. The aqueous supernatant was discarded and the organic phase dried at 60 °C under a stream of nitrogen gas. The residue was dissolved in one milliliter of 60:40 acetonitrile:water by vortexing on a mixer for ten seconds. The solution was then transferred into one milliliter HPLC sample vials. Paclitaxel was assayed at ambient temperature by HPLC with an injection volume of 20 uL and a mobile phase of acetonitrile:water:methanol 58:37:5 flowing at one milliliter per minute. The detection wavelength was 232 nm. Paclitaxel is partially converted in aqueous media to 7-epi-taxol. Paclitaxel and 7-epi-taxol have the same absorbance and so the area under the peak of the 7-epi-taxol chromatogram was added to the area of the paclitaxel peak to determine the quantity of paclitaxel being assayed. 3.3.15. Statistical analysis The standard deviation of samples was calculated from (Zar,1984) standard deviation = ((LX? - ( (IX,)2/ n)) / n)m where X is a measurement of a sample, and n is the size of the sample. 77 Two populations' parameters were compared by examining appropriate sample estimates, using the t distribution of each sample, referred to as the Mest or Student's Mest. The test statistic (t) was calculated from (Zar,1984) t = (X-/a) I sx Equation 58 where X is the mean of the sample, /u is the mean of the population, and sxis the estimate of the population variance. If the absolute value of t was larger than the critical value of t from Table B.3 in Zar (1984), then the two populations were taken to be different. The use of the Mest assumes that the sample data were from a normal population, which would be expected to result in a calculated mean from a normal distribution of means. However, the Mest is robust, in that moderate deviations from this assumption do not seriously affect the validity of the test (Zar, 1984). A variable was compared between three or more samples, using the analysis of variance (ANOVA) test. The ANOVA was reviewed by Zar (1984). The ANOVA concludes whether or not all of the tested population means are equal or not. The use of the ANOVA assumes that the sample data were from a normal population, and that the population was homogeneous. However, the ANOVA is also robust, in that even considerable deviations from these two assumptions do not seriously affect the validity of the test. Although the ANOVA concludes whether or not all of the population means are equal or not, if they are not all equal, it does not tell us how many differences there are or where the differences are located among the population means. If the ANOVA concluded that there was a difference in the population means, then the Tukey test was used to determine which means were different. The Tukey test was reviewed by Zar 78 (1984). The use of the Tukey test has the same underlying assumptions as the use of the ANOVA, that is, that the sample data were from a normal population, and the population was homogeneous. 79 4. RESULTS 4.1. The physicochemical properties of paclitaxel loaded EVA and polyurethane films 4.1.1. Thermal analysis of films DSC thermograms were obtained between -75 to 125 °C at 10 °C-min"1 for control (no drug), 5, 10, 20, and 30 % w/w paclitaxel loaded 60/40 EVA, 72/28 EVA, and polyurethane films, nonsterile and sterilized. Representative thermograms of 60/40 EVA, 72/28 EVA, and polyurethane films are shown in Figure 9. There was a broad complex endotherm from approximately -40 to 45 °C for the 60/40 EVA films, and a broad complex endotherm from - 35 to 70 °C for the 72/28 EVA films. The 60/40 EVA and 72/28 EVA films both had small, sharp endotherm peaks at approximately 37 °C. The 72/28 EVA films had a second small, sharp endotherm at approximately 65 °C. The endotherm peaks at approximately 37 °C for both 60/40 EVA and 72/28 EVA were taken to be melting peaks (Tm). The Tm of the control, 5, 10, 20, and 30 % w/w paclitaxel loaded 60/40 EVA and 72/28 films, nonsterile and sterilized, are summarized in Table 1. The Tm of the various EVA films ranged from 35.6 to 38.0 °C. EVA monomer ratio and drug loading had no observable effect on Tm. Using two-tailed Mest, there was no significant difference (p > 0.05) in the Tm between nonsterile and sterilized 60/40 EVA films nor between nonsterile and sterilized 72/28 EVA films. 80 The degree of crystallinity of the polymer for each film was calculated from the AHf determined from the DSC runs and the given value for AHj\oo% (Equation 8). The degree of crystallinity of the polymer of control, 5, 10, 20, and 30 % w/w paclitaxel 60/40 EVA and 72/28 EVA films, nonsterile and sterilized, are summarized in Table 2. The degree of crystallinity of the polymer of the various EVA films ranged from 2.14 to 4.12 %. Using two-tailed Mest, 5, 10, and 20 % w/w paclitaxel loaded, nonsterile films of 72/28 EVA, had a significantly greater (p < 0.05) degree of polymer crystallinity than those films of 60/40 EVA; similarly, control, 5, 10, and 20 % w/w paclitaxel loaded, sterilized films of 72/28 EVA, had a significantly greater degree of polymer crystallinity than those films of 60/40 EVA. No trend was observed for drug loading and polymer crystallinity. Using two-tailed Mest, control, 5, 10, and 20 % w/w paclitaxel loaded 60/40 EVA films that were nonsterile, had a significantly greater (p < 0.05) degree of polymer crystallinity than those films that were sterilized; similarly, control, 5, and 10 % w/w paclitaxel loaded 72/28 EVA films that were nonsterile, had a significantly greater degree of polymer crystallinity than those that were sterilized. No DSC thermal events were observed for any polyurethane films. Representative thermogravimetric analysis (TGA) thermograms of 60/40 EVA, 72/28 EVA, and polyurethane films are shown in Figure 10. Thermogravimetric analysis showed that there was no weight loss on heating for any of the films until at temperatures greater than 300 °C. Boiling and blackening of the films accompanied this weight loss. No thermal events were attributable to paclitaxel in any of the films. 81 -14 -100 -50 50 100 Temperature (°C) 82 Table 1. Melting temperature peak (Tm) of 5, 10, and 30 % w/w paclitaxel loaded 60/40 E V A and 72/28 E V A films, nonsterile and sterilized. Values (°C) are the mean ± standard deviation of the measurements of five samples. Tm (°C)a Polymer Paclitaxel (%) 0 5 10 20 30 60/40 EVA Nonsterile 36.9 + 0.71"4 37.0+ 0.51"4 37.8 + 0.73'4 37.9 + 0.63'4 37.6 + 0.72"4 SterUe 37.7 + 0.63'4 37.9 ± 0 . 2 4 38.0 ± 0 . 9 4 36.8 ± 0 . 3 M 36.7 ± 1.3 M 72/28 EVA Nonsterile 3 6 . 6 ± 0 . 8 M 36.9 + 0.31"4 35.6+ 0.51 36.2+ 0.11'2 36.4 ±0.2' ' 3 Sterile 36.8+ 0.21"4 37.3 + 0.42"4 36.4+ 0.31"3 36.4+ 0.41"3 36.6 + 0.41-4 a Tm was taken to be the small endofhermic peak within a broad complex endotherm. 1 - 4 Using A N O V A and Tukey tests, there was no statistical difference in the means of samples denoted with the same superscript number. Some means have a range of 1 3 superscript numbers, which then includes all numbers in the range. For example, represents ' , 2 ' 3 . Table 2. Degree of crystallinity of 5, 10, and 30 % w/w paclitaxel loaded 60/40 E V A and 72/28 E V A films, nonsterile and sterilized. Values (%) are the mean ± standard deviation of the measurements of five samples. Degree of crystallinity (% a Polymer Paclitaxel (%) 0 5 10 20 30 60/40 EVA Nonsterile 3.10+0.391"3 2.54+0.561"3 2.58+0.21'"3 3.39+0.451-4 3.64+0.732"4 Sterile 2.51+0.371"4 2.1410.151 2.24+0.231,2 2.80+0.21M 2.86+1.911-4 72/28 EVA Nonsterile 3.74+0.573'4 3.73±0.61 3' 4 4.09+0.464 3.93+0.593'4 2.47+0.361-3 Sterile 2.73±0.46 M 3.07+0.681"4 3.55+0.41M 4.12+0.444 2.70+0.471"4 The degree of crystallinity was determined from Equation 8: %crystallinity = AH V / i o o % xl00% '° J 1 - 4 Using A N O V A and Tukey tests, there was no statistical difference in the means of samples denoted with the same superscript number. Some means have a range of superscript numbers, which then includes all numbers in the range. For example, 1 - 3 represents ''2'3. 84 o 4.1.2. X-ray Figure 11 shows representative diffraction patterns of nonsterile 60/40 EVA, 72/28 EVA, and polyurethane films. Similar data were also obtained for 60/40 EVA, 72/28 EVA and polyurethane films with and without paclitaxel, unsterilized and sterilized. These films showed no distinct peaks, rather, a single, broad peak was observed for all samples. Figure 11. XRD patterns of films prepared from A) 60/40 EVA, B) 72/28 EVA, and polyurethane, scanned at 5° 29-min"1 at increments of 0.02° 20. 9000 , — — -j 0 10 20 30 40 50 60 Diffraction angle (degrees 2-theta) 85 4.1.3. Molecular weight determination of polymers The [r|] values for 60/40 EVA, 72/28 EVA, and polyurethane were determined by viscometry. These values were required for the GPC universal calibration curve. The MWs of the various films for drug release studies were then determined by GPC with reference to the GPC universal calibration curve. 4.1.3.1. Viscometry of films Huggins and Kramer plots for 60/40 EVA, 72/28 EVA, and polyurethane are shown in Figure 12. Table 3 summarizes the intrinsic viscosity data calculated using the Kraemer (1938), Huggins (1942), and Solomon and Ciuta (1962) equations (Equation 5, Equation 4, and Equation 6, respectively). The mean of the Huggins fl/nj^and Kraemer ([r|]it) intrinsic viscosities ([T|]HK), was similar to the Solomon and Ciuta viscosity ([r|]sc), for each polystyrene MW standard. The differences varied from 0.004 dL-g"1 for the 17 500 g -mol"1 standard (1.3 % difference) to 0.062 dL-g"1 for the 233 000 g-mol"1 standard (5.5 % difference). As the intrinsic viscosity results were similar for each polystyrene MW standard for each method of calculation, only the values of [T|]HK for 60/40 EVA, 72/28 EVA, and polyurethane were determined (Figure 10 and Table 3). The values of [T)]HK for the various polymers increased in the following order: 72/28 EVA < 60/40 EVA < polyurethane. 86 Figure 12. 0;5 Huggins (Equation 4) and Kraemer (Equation 5) plots of viscosity rj data for A) polyurethane, B) 60/40 EVA, and C) 72/28 EVA in chloroform at 25 ± 0.5 °C. Y-intercept values of each line are the respective Huggins ( [ T ] ] H ) or Kraemer ([r|]K) intrinsic viscosity of the polymer. • Huggins • Kraemer 0.2 0.4 0.6 Concentration (% w/v) 0.8 Regression analysis parameters (y = mx + b) and R : Huggins plot for polyurethane: Kraemer plot for polyurethane: Huggins plot for 60/40 EVA: Kraemer plot for 60/40 EVA: Huggins plot for 72/28 EVA: Kraemer plot for 72/28 EVA: y= 1.38x+ 1.47 and R^ = 0.979 y = -0.148x + 1.53 and R2= 0.587 y = 0.315x + 0.943 and R2= 1.000 y = -0.128x + 0.941 and R2= 1.000 y = 0.064x + 0.687 and R2= 0.994 y = -0.115x + 0.672 and R2= 1.000 87 Table 3. Intrinsic viscosity data for polystyrene standards, 60/40 EVA, 72/28 EVA, and polyurethane. Standard polystyrene values represented are a mean calculated from three measurements ± standard deviation. Intrinsic viscosity ± standard deviation (dLg1) Polymer Huggins ([71H) Kraemer Solomon And Ciuta ([Vhc) Mean Huggins and Kraemer ( M H K ) Polystyrene MW 17 500 g mol"1 0.330±0.032 0.305±0.026 0.313±0.028 0.317±0.029 Polystyrene MW 50 000 g mol"1 0.347±0.p024 0.328±0.021 0.3343±0.022 0.337±0.023 Polystyrene MW 100 000 g mol"1 0.509±0.033 0.469±0.028 0.482±0.030 0.489±0.035 Polystyrene MW 233 000 g mol"1 1.361±0.093 1.027±0.048 1.128±0.021 1.19±0.19 Polystyrene MW 300 000 g mol"1 1.486±0.024 1.279±0.026 1.345±0.025 1.38±0.12 Polystyrene MW 600 000 g mol"1 2.044±0.075 1.856±0.051 1.916±0.059 1.95±0.12 60/40 EVA 0.943 0.941 — 0.942 72/28 EVA 0.687 0.672 — 0.680 Polyurethane 1.472 1.531 — 1.502 4.1.3.2. Gel permeation chromatography of films The universal GPC calibration curve based on polystyrene standards is shown in Figure 13. Representative GPC elution profiles of nonsterile and sterilized films prepared from 60/40 EVA, 72/28 EVA, and polyurethane are shown in Figure 14. The 88 A /GPC values for the nonsterile 60/40 EVA, 72/28 EVA, and polyurethane films are given in Table 4.. The sterilized 60/40 EVA and 72/28 EVA films' peaks were broad (Figure 12, figures labelled "b"), retention times ranging from approximately four to seven minutes, which made it difficult to determine their MQPC - The nonsterile and sterilized 60/40 EVA and 72/28 EVA films both had shoulder peaks at approximately 4.1 min, with the shoulder peak being more pronounced for the sterilized films. The shoulder peaks at 4.1 min were of greater M G P C than the standards used in the universal calibration curve. There was no observed effect of paclitaxel loading on the MQPC of any of the polymer types. Table 4. The M G P C of 60/40 EVA, 72/28 EVA, and polyurethane from nonsterile films. Film MGpc ±S. D. (gmol1) 60/40 EVA 64 200 ± 7 200a 72/28 EVA 23 300 ± 1 500 Polyurethane 126 000 ± 14 000 a Using ANOVA and Tukey tests, the M G P C of the various nonsterile films increased significantly from 72/28 EVA to 60/40 EVA to polyurethane. 89 Figure 13. GPC universal calibration curve of polystyrene molecular weight standards. 6 , 5.5 -3.5 J, , , = , , : , 1 4.5 5 5.5 6 6.5 7 7.5 8 Retention time (min) 2 Regression analysis parameters (y = mx + b) and R : y = -0.714x + 9.15 and R2= 0.999 GPC elution profiles of a) nonsterile and b) sterilized films prepared from A) 60/40 EVA, B) 72/28 EVA, and C) polyurethane. 91 4.1.4. Rheology of films The rheological properties of selected films used in in vitro release studies were determined by dynamic mechanical analysis (DMA). These properties included the complex viscosity (rf), storage modulus (G'), loss modulus (G"), and complex relaxation time (A*). Representative rheology plots for films prepared from 60/40 EVA, 72/28 EVA, and polyurethane are shown in Figure 15. Using ANOVA and Tukey tests, the values of rf, G \ and G" observed at a frequency of 1 rad-s1, of the various nonsterile films (without drug), increased significantly from 72/28 EVA films to 60/40 EVA films to polyurethane films. Using two-tailed t-test, sterilized films made up with 60/40 EVA and 72/28 EVA, had significantly greater (p < 0.05) values of rf, G \ and G", than nonsterile 60/40 EVA and 72/28 EVA films, respectively, when observed at a frequency of 1 rad-s"1. Using two-tailed Mest, 30 % w/w paclitaxel loaded sterilized 60/40 EVA and 72/28 EVA films had significantly greater (p < 0.05) values of values of rf, G \ and G", than non-drug loaded sterilized 60/40 EVA and 72/28 EVA films, respectively, when observed at a frequency of 1 rad-s"1. Neither sterilization nor paclitaxel loading had any observed effect on the values of rf, G \ and G" for polyurethane films. Using Equation 27, the A* was determined from the frequency at which G' was equal to G"(that is, the frequency at which the G' and the G " plots cross in Figure 15). The A* data for the various films are summarized in Table 5. Using two-tailed Mest, sterilized 60/40 EVA films had significantly greater (p < 0.05) A* values than nonsterile 60/40 EVA films, 92 while sterilized polyurethane films had significantly lower (p < 0.05) A* values than nonsterile polyurethane films. Table 5. Characteristic complex relaxation time (A*) for paclitaxel loaded 60/40 EVA, 72/28 EVA, and polyurethane films. A* Sterilization Paclitaxel (%) 72/28 EVA 60/40 EVA Polyurethane Unsterilized 0 < 0.0001 < 0.0001 -0.21 5 < 0.0001 < 0.0001 -0.82 Sterilized 0 < 0.0001 -32 - 0.052 30 < 0.0001 -82 -0.032 G', G " (dyne/cm2), and TJ* (g/cm.s) G', G" (dyne/cm2), andr|* (g/cm.s) •3 C i i I •3 I o H - O O o o o H - o o o o H - o o o o o o o o o o o _1_ • L -•2 o o • • » • • • • • • • a • • • • • • • • • • i— o o 1 — 0 0 0 i— o o o o i— o o o o o o o o o o o o o o i— o o H - o o o H - o o o o •-' o o o o o o o o o o o o o o >— o o o o o o o o o o o o o o o o o o o o o o H - O i— o o H - o o o >— o o o o H - o o o o o o o o o o o o o Cd h- o >-» O O i— O O O O O O O O o o o o o o o £1 do' c a (71 m n CO fD 3 f f l fD p B 3 Cu •n cu fD ft o o 3 * fD H 3 * fD o ^ o co 8 . ! * o o 3 o 3 cn r-t-fD 3. 3 * 5* 3 fD SI s CX) o O ft • T3 to fD Cu 3> O 3 ^0 0 0 -a £ ° — & 3 •-»- CO CU Q , fD ^ Cu ^ 3 £ N> N5 oo m > 3 o o 3 Cu e6 94 4.1.5. Thickness of films for drug release studies The thickness of paclitaxel loaded 60/40 EVA, 72/28 EVA, and polyurethane films, nonsterile and sterilized, are given in Table 6. The thickness of the various films ranged from 112 - 161 um for polyurethane, 123 - 151 um for 60/40 EVA, and 162 - 209 um for 72/28 EVA. Table 6. The thickness (h) of 5, 10, and 30 % w/w paclitaxel loaded 60/40 EVA, 72/28 EVA, and polyurethane films, nonsterile and sterilized. Values are the mean ± standard deviation of the measurement of five samples. h (um) Polymer Paclitaxel (%) 0 5 10 20 30 Poly- Unsterile 161 ± 4 1" 6 129 ± 1 7 1 3 1 3 1 ± 2 5 1 3 1 1 4 ± 121 123 ± 10 1 ' 2 urethane Sterile 143 ± 3 0 M 1 3 7 ± 2 5 M 141 ± 1 2 M 122 ± 2 6 u 1 1 2 ± 161 60/40 Unsterile 151 ± l l 1 " 5 149 ± IO 1" 5 140 ± 4 M 131 ± 7 1" 3 130 ± 61"3 E V A Sterile 151 ± 8 1 ' 5 143 ± 7 M 138 ± l l 1 " 4 123 ± 8 1 ' 2 129 ± 11 1 - 3 72/28 Unsterile 185 ± 20 4" 6 178 ± 143"6 166 ± 22 2" 6 170 ± 12 2" 6 162 ± 2 4 ' " 6 E V A Sterile 187 ± 41 4 " 6 209 ± 5 9 6 154 ± 2 5 1 - 5 197 ± 18 5 ' 6 175 ± 173"6 1 6 Using ANOVA and Tukey tests, there was no statistical difference in the means of samples denoted with the same superscript number. Some means have a range of superscript numbers, which then includes all numbers in the range. For example, 1 4 12 3 4 represents ' ' ' . 95 4.1.6. Visual estimation of the solubility of paclitaxel in EVA and polyurethane Table 7 summarizes the solubility of paclitaxel in 60/40 EVA and polyurethane as visually determined by the opacity of the films. 60/40 EVA films containing 0.1 to 3 % w/w paclitaxel were transparent whereas films containing 4 to 30 % w/w paclitaxel were opaque. An opaque film was taken to be a saturated solution of paclitaxel in the polymer. The opacity of the 4 to 30 % w/w paclitaxel loaded 60/40 EVA films increased with an increase in drug loading. There was no observed opacity of the polyurethane films loaded with 0.1 to 30 % w/w paclitaxel. Therefore the solubility of paclitaxel in the polymers (Cs-) was estimated to be 3 to 4 % w/w for 60/40 EVA and greater than 30 % w/w for polyurethane. Table 7. Solubility of paclitaxel in 60/40 EVA and polyurethane determined using various methods. Polymer Type 60/40 EVA Polyurethane 0.38 0.36 r b 0.933 ± 0.09 0.59 ±0.41 CS"C' 3-4 >30 a Cj - was determined from diffusion cell studies and Equation 39 b Cr" was determined from drug release studies, and Equation 57. 0 Cs-" was determined from the visual estimation method. 96 4.1.7. Paclitaxel calibration curves The linearity of all sixteen paclitaxel calibration curves for paclitaxel (four a day for four days), expressed as R2, was 0.9999. Representative values of the accuracy of injections of paclitaxel at each concentration on each day are shown in Table 8. Values of coefficient of variation (CV) were observed to be less than 15% for all concentrations of paclitaxel (0.1 to 50 ug-mL"'). Bias was observed to be less than 15% for all concentrations. Representative values of intraassay precision and a summary of interassay precision for the paclitaxel standard curve are shown in Table 9. Values of intraassay precision were observed to have a CV of less than 15% for all concentrations. Values of interassay precision were observed to have a CV of less than 15% for all concentrations, except for the lowest concentration (0.1 lag-mL"1), which had a CV of 20.14%. 97 Table 8. Accuracy of paclitaxel calibration curves. These curves were determined on day one of four days. Data for days two, three, and four are not shown. Paclitaxel Predicted concentration value8 C V b (%) Biasc (%) Gig-mL"1) (ug-mL"1) 0.1 0.13 2.86 1.94 0.2 0.23 2.21 4.98 0.5 0.59 13.62 12.11 1 1.01 1.41 0.50 2 2.06 0.70 0.80 5 5.75 13.28 13.13 10 10.22 0.14 0.78 20 20.46 0.79 0.27 50 48.94 1.05 0.54 a Predicted value is the mean of concentration values calculated from three curves on each day. b CV (coefficient of variation) is the quotient of the standard deviation by the mean, and is expressed as a percentage. A value of less than 15% is taken to be acceptable accuracy, except at the limit of quantitation, where it should not deviate by more than 20% (Shah, 1992). 0 Bias is the ratio of the deviation of predicted value from the actual concentration measured, and is expressed as a percentage. A value of less than 15% is taken to be acceptable accuracy, except at the limit of quantitation, where it should not deviate by more than 20% (Shah, 1992). 98 Table 9. Representative intraassay precision and summary of interassay precision of paclitaxel calibration curves. Paclitaxel concentration (Ug-mL"1) Day l a All days Peak areab CVC (%) Peak area CV (%) 0.1 10 453 2.53 8211 20.14 0.2 18 050 3.01 15 688 11.40 0.5 48 553 12.65 44 856 13.94 1 80 325 1.18 75 575 10.91 2 165 365 0.70 148 563 12.58 5 476 576 12.74 437 125 14.20 10 813 756 0.41 811 563 1.13 20 1 631 258 0.66 1 628 750 0.79 50 3 899 758 0.90 3 987 333 1.69 a Day 1 is representative of one of four days (four curves for each day). b Peak area is the mean calculated from four standards at each concentration each day. c C V (coefficient of variation) is the quotient of the standard deviation by the mean, and is expressed as a percentage. A value of less than 15% is taken to be acceptable accuracy, except at the limit of quantitation, where it should not deviate by more than 20% (Shah, 1992). 99 4.2. The effect of polymer type on the diffusion of paclitaxel through EVA and polyurethane films In vitro cumulative diffusion of paclitaxel through 60/40 EVA and polyurethane films with time is shown in Figure 16. The diffusion parameters obtained are summarized in Table 10. Sample calculations for the diffusion parameters are shown in Appendix I. The theoretical time lag (ti) was determined by extrapolating the linear regression analysis line of the four terminal points to the x-axis. The diffusion coefficient (D) was calculated from the ti with the film thickness (h) in Equation 43. The permeability coefficient (P) was calculated from the slopes of the linear regression analysis lines (Equation 41). The partition coefficient (K) was calculated from the h, P, and D in Equation 36. The solubility of paclitaxel in each polymer (Cs) was calculated from the P and the solubility of paclitaxel in the donor compartment (CJ) in Equation 39. 100 Figure 16. In vitro cumulative diffusion of paclitaxel through A) 60/40 EVA and B) polyurethane films. The points on the graph are the mean of three samples, and the error bars represent one standard deviation. The lines are the linear regression analysis lines of the four terminal points, extrapolated to the x-axis. Regression analysis parameters (y = mx + b) and R 604/40 EVA: y = 10.2x - 19.86 and R2= 0.992 Polyurethane: y = 7.09x - 17.09 and R2 = 0.994 101 Table 10. Coefficient values obtained from the cumulative diffusion of paclitaxel through 60/40 EVA and polyurethane films. Film Type 60/40 EVA Polyurethane ha (urn) 148 ± 6 135 ± 10 hid) 1.95 2.41 D (cm2 s"1) 2.16 x 10"10 1.46 x 10"10 S (mm2) 211 Cd (ug mL"1) 1.45 ±0.07 P (cms1) 3.85 x 10"5 2.67 x 10"5 K 2640 2480 Cs> (% w/v) 0.383 0.360 a Values of thickness are the mean ± standard deviation of the measurement of three films. 4.3. The effect of monomer and polymer type on the release of paclitaxel from EVA and polyurethane films In vitro release profiles of paclitaxel from films prepared from 60/40 EVA, 72/28 EVA, and polyurethane are shown in Figure 17. The release of paclitaxel from all films was biphasic. Paclitaxel release was greatest from 60/40 EVA films for the first 5 to 14 days, from 72/28 EVA films for the first 19 days, and from polyurethane films for the first 9 days. The initial rapid release phase for paclitaxel from all films was followed by a slower phase of release. Based on a solubility value of paclitaxel in the release medium of 1.45 ± 0.07 ug-mL"1 (Table 9), it is evident that concentrations near saturation were encountered in the first several days of the study. Using ANOVA and Tukey tests, 5 % w/w drug loaded 72/28 EVA films released paclitaxel to a significantly greater (p < 0.05) extent than 5 % w/w loaded polyurethane films, and 30 % w/w drug loaded 60/40 EVA 102 films released paclitaxel to a significantly greater extent than 30 % w/w drug loaded 72/28 EVA films. Between 26.8 and 60.7 % of the total paclitaxel was released over the forty-two days of the release study. 4.4. The effect of drug loading on the release of paclitaxel from EVA and polyurethane films In vitro release profiles of paclitaxel from 5, 10, and 30 % w/w drug loaded EVA and polyurethane films are shown in Figure 17. Using ANOVA and Tukey test, and comparing films of the same polymer type, increased drug loading levels from 5 to 10 to 30 % w/w resulted in a significantly greater cumulative release of paclitaxel (115 - 158, 208 - 270, and 458 - 647 ug for 5, 10, and 30 % w/w paclitaxel loaded films, respectively). Cumulative paclitaxel release (p:g) 5 do" s fD ft> o, 3 fO a Cu o o o o o o o o o o o o o o o o o t o o o o o o o o o o o o o o o o o o o o o o o o o o CD ^ t o w j i i y i ^ v i o o o o o o o o o o o o o o o o u > o o o O 3 b i 5 o a 0 s o * i | co • & 1 d £ sL « ft <t M l l - l B 3 n B 05 P EL SS ^ fD "* a o 3 o g 2 P O co 5 . ^ P Cu fD fD fD fD Cu o p " fD 3 t 3 ' P w o S 3 P ^ | 3 • " ^ C O OJ fD O ^3 P fD Cu SP o 3 ON O ^ <; w > T J -r | 3 C O —1 O ^ 3 N> r-l- 00 & HH fD m < P > e» a. fD n 3" fD eoi 104 4.5. Modeling of drug release from paclitaxel loaded EVA and polyurethane films According to the kinetics described for diffusion controlled release from a saturated non-degradable matrix by Higuchi (Higuchi, 1960) (Equation 56), the cumulative amount of drug released is directly proportional to the square root of time. In vitro cumulative release of paclitaxel as a function of the square root of time, from 5, 10, and 30 % w/w drug loaded EVA and polyurethane films, is shown in Figure 18. For all films, a linear relationship was found between the cumulative amount of paclitaxel released and the square root of time for the first several days (for the first 5-9, 10 - 15, and 9 days for 60/40 EVA, 72/28 EVA, and polyurethane films, respectively). CA — i — W i o vo -a - J X X ' + X ta 3 CL SO II II o P o V O g V O oo g oo U) ^ ^ v< CA oo bv l/i x x + 4^ O s CL 4L X + OO CL CL ?o ?d ?d vo o o V Q vo vo >S V O V O 0 0 +>. o v< <<; U> CA c/i oo 4^ X + p CA 4^ O CL 4^ X O to X + CA CA o ; , o vo —J oo oo to o V O V O vo CA C O 5' 13 s T3 8 CD CD i-t co a. 70 fD fD CL fD s fD CO & >-t° fD O vT §• o <: fD Cumulative paclitaxe release (ug) M U > U Ot ^ 00 O O O O O O O o o o . o o o o o 3 g to o o o o o o o ~ o o o o o o o o JS -to • - t o u i ^ u i C A ^ i c e o o o o o o o ~ o o o o o o o o o f " fD o fD do e n 00 5- 3 3 fD so g 5 So o «<* ^ 3 - co 3 JO fD 3 co 3 o U |-b fD (-»• CO fD, 05 If fD ^ 3 — fD jo co H 3^ fD Eft 3 o 3 3 * o 3 O 3 * fD CT Sa I-I co i-t fD ^ i-t fD co fD 3 8 <» CL *=t ~ <„ ^ CL CL CL fD <_ so' c-t-o* 3 H 3 -fD P Eft § 3 0 s fD ? £0 > |-t 5? fD EV 3 S > fD -—' SOI 106 According to the kinetics described for diffusion controlled release from a saturated non-degradable matrix by Higuchi (Higuchi, 1960) (Equation 56), the cumulative amount of drug released at a given time point is directly proportional to the square root of the concentration of the drug loading. The cumulative release at 2.56 d was chosen as a given time point to model the data, as this time point was approximately in the middle of the time period of the linear cumulative release of paclitaxel as a function of the square root of time. In vitro cumulative release of paclitaxel at 2.56 d as a function of the square root of the concentration of the drug loading, from 5, 10, and 30 % w/w drug loaded EVA and polyurethane films, is shown in Figure 19. For each polymer film, a linear relationship was found between the cumulative amount of paclitaxel released at 2.56 d and the square root of the concentration of the drug loading. 107 Figure 19. In vitro cumulative release of paclitaxel (in at 2.559 d as a function of the square root of loading concentration from films prepared from A) 60/40 EVA, B) 72/28 EVA, and C) polyurethane. The films were loaded with 5, 10, and 30 % w/w paclitaxel. The points on the graph are the mean of five samples, and the error bars represent one standard deviation. The lines on the graphs are the linear regression analysis lines of the points. 450 -j , 400 A 50 A o -! 1 1 — - — i 1 ' 1 1 0 1 2 3 4 5 6 7 Square root of paclitaxel loading concentration (square root of (% w/w)) Regression analysis parameters (y = mx + b) and R : 60/40 EVA: y = 74.1 x - 61.0 and R2 = 1.000 Polyurethane: y = 63.8x - 80.6 and R2 = 1.000 72/28 EVA: y = 41.1x - 24.1 and R2 = 0.999 108 5. DISCUSSION 5.1. The physicochemical properties of paclitaxel loaded EVA and polyurethane films The physical state of both drug and polymer in a polymeric matrix play a role in the drug release process (Bernabei, 1982). The physical state of paclitaxel and the polymers in the various films was investigated by DSC (Figure 9) and TGA (Figure 10). No thermal events were observed for paclitaxel in any of the films. This suggested that paclitaxel was miscible with the polymer matrices or existed as an amorphous solid within the polymer matrices. It may also be that if some crystalline paclitaxel were present, it may have existed in too small an amount for thermal events due to the paclitaxel to have been detected. Polymer crystallinity may play a role in the release of drug from a polymer matrix. As the crystallinity of a polymer matrix is increased, the matrix becomes less permeable to drugs (Pitt, 1990). The degree of crystallinity of a polymer matrix may affect the diffusion of drug through and release of drug from the matrix. The lack of any observed thermal events for polyurethane films suggest that these films were completely amorphous. The Tg for these films could not be detected. The broad, complex, DSC endotherms observed for 60/40 EVA and 72/28 EVA films (Figure 9) may be due, in part, to the effects of solidifying the films from solution in DCM. After both types of films were melted, their thermal history was erased. Upon cooling from the melt and immediately reheating, broad endotherms were observed; however, the small, sharp 109 peaks at 37 °C for both film types and at 65 °C for 72/28 EVA, were not observed. The peaks observed at 37 °C for both film types may have been due to secondary crystallization. The degree of polymer crystallinity determined for the various paclitaxel loaded EVA films ranged from 2.5 - 3.6 % and from 2.1 - 2.8 % for nonsterile and sterilized 60/40 EVA films, respectively, and from 2.5 - 4.1 % and from 2.7 - 4.1 % for nonsterile and sterilized 72/28 EVA films, respectively. The degree of polymer crystallinity was determined to be greater for some of the paclitaxel loaded 72/28 EVA films than for similar 60/40 EVA films (Table 2). The greater crystallinity of the EVA with the larger ethylene to vinyl acetate ratio is expected as the ethylene segments may form crystalline regions whereas the vinyl acetate segments are completely amorphous (Salyer, 1971). Salyer (Salyer, 1971) determined 62/38 EVA and 80/20 EVA to be amorphous and 19.9 % crystalline, respectively. The degree of polymer crystallinity was determined to be greater for some of the nonsterile, paclitaxel loaded 60/40 EVA and 72/28 EVA films, than for similar sterilized films (Table 2). This suggested that y-irradiation of both types of EVA films may have resulted in cross-linking, which would have resulted in less secondary crystallization, and therefore decreased observed crystallinity. The degree of crystallinities of the EVA films (Table 2) were determined assuming that the small DSC peaks observed at 37 °C (Figure 9) were due to melting. The DSC endotherms observed, from approximately -40 to 45 °C for the 60/40 EVA films, and from approximately - 35 to 70 °C for the 72/28 EVA films, were broad and 110 complex. There may be some error in attributing the small peaks at 37 °C to melting of crystallites. There was no observed weight loss with TGA heating for any of the film types until temperatures greater than 300 °C were reached (Figure 10). This suggested that the broad, complex endotherms observed in the DSC runs for 60/40 EVA and 72/28 EVA films (Figure 9) were not due to loss of residual solvent. The boiling and blackening of the films associated with weight loss at temperatures greater than 300 °C suggested that this weight loss was due to degradation of the films. The physical state of paclitaxel and the polymers in the various films was also investigated by XRD (Figure 11). No distinct peaks were observed in the X-ray diffraction patterns for any of the film types. Rather, the X-ray diffraction patterns of these films showed a diffuse pattern (an "amorphous halo") which is characteristic of a disordered structure (Hearle, 1982). These data confirmed the DSC results that showed that polyurethane films were completely amorphous and that 60/40 EVA and 72/28 EVA films had low degrees of crystallinity. Polymer MW may influence the release of drug from a polymer matrix. As the MW of a polymer matrix is increased, the matrix becomes less permeable to drugs, due to an increased density and an increased glass transition temperature (Omelczuk, 1992). The MWs of the polymers used in the various films for drug release studies were determined by GPC (Figure 14) with reference to the GPC universal calibration curve (Figure 13). The nonsterile and sterilized 60/40 EVA and 72/28 EVA films both had shoulder peaks at approximately 4.1 min, with the shoulder peak being more pronounced I l l for the sterilized films. The shoulder peaks at 4.1 min were of greater M G P C than the standards used in the universal calibration curve. That is, the shoulder peaks had a M G P C > 233 000 g-mol"1. Although the M G P C cannot be extrapolated accurately outside of the range of standards used in the universal calibration curve, the M G P C of the shoulder peaks was estimated by extrapolation to be greater than 2 000 000 g-mol"1. These data suggested that y-irradiation of EVA resulted in cross-linking of polymer chains. In the presence of oxygen, y-irradiation produces free radicals in polymer chains (Sintzel, 1997), which then covalently bond to each other, resulting in a cross-linked polymer. The sterilized polyurethane films had a lower MGpc than the unsterilized films. This suggested that the polyurethane films underwent chain scission when sterilized by y-irradiation. There was no observed effect of paclitaxel loading on the MGpc of any of the polymer types. Preferred design features of a perivascular wrap for drug delivery to the adventitia, include elasticity or flexibility. A higher MW of a polymer may result in a lower amount of chain slippage under shear stress, which would be expected to be observed as greater values of G \ G", 77*, and A*. The rheological properties of selected films were determined by DMA, by determining the values of G \ G", and 77* with a strain sweep through a range of frequencies from 0.1 to 100 rad-s"1. The values of A* were determined from the frequency at which G' was equal to G ". At a frequency of 1 rad-s"1, the 60/40 EVA films had higher 77*, G \ and G" values than did the 72/28 EVA films (Figure 15); the 60/40 EVA films also had higher values of X* than did the 72/28 EVA films (Table 6). This may have been due in a small part, to the higher ratio of 112 ethylene to vinyl acetate monomers of 72/28 EVA, producing suffer polymer chains than 60/40 EVA. Thus, the 72/28 EVA polymer chains would be expected to have fewer entanglements than the 60/40 EVA polymer chains. A smaller number of entanglements would result in an increase in chain slippage under shear stress. However, the difference in rheological properties between the two types of EVA, most likely may have been due to the lower MW of the 72/28 EVA. The lower MW of 72/28 EVA chains, would also be expected to produce fewer entanglements than the 60/40 EVA chains. Qualitatively, upon handling, the 72/28 EVA films did not seem as flexible as the 60/40 EVA and polyurethane films. The 60/40 EVA films were also tacky and would adhere to themselves if folded over. y-irradiation of polymers may result in chain scission (decrease in MW), and/or crosslinking (increase in MW), and a decrease in crystallinity. There may be a change in the mechanical properties of a polymer when y-irradiated, due to chain scission and/or crosslinking (Sen, 1995). With sterilization, there was an increase in the values of G\ G" and TJ*, for 60/40 EVA and 72/28 EVA films (Figure 15). A higher MW of a polymer may result in a lower amount of chain slippage under shear stress, which would be expected to be observed as greater values of G \ G", tj*, and A*. These data suggested that y-irradiation of EVA films may have resulted in cross-linking of the polymer chains, in agreement with the results from the GPC that suggested that the sterilized EVA films had more cross-linked polymer than the nonsterilized EVA films. 113 The strain sweep of the sterilized, non-drug loaded 60/40 EVA films, showed that G" was greater than G' for the entire range of frequencies (data not shown). That is, the films produced a more viscous than elastic response, and this may have been due to a weak interaction between the polymer chains (Hagan, 1995). However, the strain sweep of the sterilized, 30 % w/w paclitaxel loaded 60/40 EVA films, showed that G' increased and became greater than G" as the frequency was increased (Figure 15E). The calculated value of A* for 30 % w/w drug loaded, sterilized 60/40 EVA films, was thus greater than 1, while for non-drug loaded, sterilized 60/40 EVA films, it was less than 1. That is, the sterilized 60/40 EVA films loaded with drug produced an elastic response, while the sterilized 60/40 EVA films produced a more viscous response. The elastic response of the sterilized, paclitaxel loaded 60/40 EVA films may have been due to a strong interaction between the polymer chains (Hagan, 1995). Therefore, incorporation of drug into the sterile 60/40 EVA films apparently caused an increase in the strength of polymer chain interactions. One possible explanation for this is that paclitaxel resulted in the formation of additional bonds within the matrices. Sterilized polyurethane films had lower A* values than nonsterile polyurethane films. The lower A* values of the sterilized films indicates a decrease in the elasticity and an increase in the chain slippage of the polyurethane films. These data suggested that y-irradiation of polyurethane films may have resulted in chain scission, in agreement with the results from the GPC that also suggested that the sterilized polyurethane films had undergone chain scission. 114 5.2. Experimental variables in paclitaxel diffusion and release studies with EVA and polyurethane films Adsorption of drug to, or absorption of drug into, containers used in diffusion or release studies may result in loss of drug and lower measured drug concentrations. Song (Song, 1996) showed that after 19 h storage of a paclitaxel solution in 1% methanol in 1.5 mL glass vials, that the concentration of the paclitaxel solution decreased to 40% of the original paclitaxel concentration. A methanol wash failed to recover any of the paclitaxel from the glass vials. The authors suggested that their data indicated that there was nonspecific adsorption of paclitaxel to glass surfaces. However, they also showed that there was no loss of paclitaxel from solution in the presence of fetal bovine serum, due to paclitaxel binding to serum proteins. Previous studies in our laboratory have shown that there was no significant loss of paclitaxel via adsorption on, or absorption into, containers. The PBSA used as the media in the previous studies and in the paclitaxel diffusion and release studies in this work contained 0.4 % w/v albumin. It is suggested that the lack of significant adsorption of the drug to the diffusion cells or release study containers may be attributed to the binding of paclitaxel to albumin in the release medium. Pseudo first-order kinetic degradation of paclitaxel in water at pH 7 to 7-epi-taxol and to other degradation products, including baccatin III, deacetylbaccatin III, baccatin V , and their 7-epimers, has been reported (Kerns, 1994; Dordunoo, 1996). Paclitaxel underwent partial epimerization in PBS at pH 9 to 7-epi-taxol and vice versa. The major paclitaxel derivative in cell culture medium was 7-epi-taxol. However, the epimerization 115 was slower in PBS than in cell culture medium (Ringel, 1987). Because 7-epi-taxol has a lower chromatographic polarity than paclitaxel, the 7-epimer elutes after paclitaxel from the reverse-phase HPLC column (Rizzo, 1990). Paclitaxel was found in this work to elute from the HPLC column at approximately 2.5 min, and 7-epi-taxol at approximately 3.5 min. Previous studies in our laboratory showed that paclitaxel and 7-epi-taxol have the same absorbance. To determine the quantity of all the paclitaxel released in this work, the area under the 7-epi-taxol peak was added to the area of the paclitaxel peak, from the HPLC chromatograms. According to the kinetics described for diffusion controlled release from a saturated non-degradable matrix by Higuchi (1960) (Equation 56), the total amount of drug released is proportional to the square root of the solubility of the drug in the matrix. The observed solubility of paclitaxel in 60/40 EVA ranged from approximately 0.4 % w/w (Cy) to 1 % w/w (Cs ••) to 3 - 4 % w/w (Csy) for the three methods of determination (Table 6). The observed solubility of paclitaxel in polyurethane ranged from approximately 0.4 % w/w (Cy) to 0.6 % w/w (Cs><) to greater than 30 % w/w (Cs-0- The values for Cs- and Cs- for the two polymers are in reasonable agreement, but are clearly lower than the visual estimate of drug solubility in the matrices, given by Cs-. The values for Cs- and Cs- are based on the assumption that the polymer film does not undergo swelling. Previous work in our laboratory showed that 60/40 EVA with or without paclitaxel did not swell in PBS A (unpublished data). However, the polyurethane manufacturer's literature indicates that polyurethane is hygroscopic and can swell approximately 2 to 3 % in aqueous media (Thermedics Inc. company literature, 1995). The values for Cs- are based on the Higuchi model, which assumes that the films are 116 saturated with drug (that is, that A » C$), and that the drug is homogeneously distributed throughout the matrices. These assumptions may not hold for polyurethane at 30 % w/w paclitaxel loading or for 60/40 EVA and 72/28 EVA at low loadings. If it is assumed that the visual estimation method gave a more reliable measure of the solubility of paclitaxel in 60/40 EVA, that is, a value of 3 - 4 % w/w, then a loading or total initial content (A in the Higuchi model) of 5% w/w would not have been much greater than the solubility. These paclitaxel solubility data suggest that paclitaxel existed mostly as an amorphous and separate phase from the polymer within the polymer matrix for the 60/40 EVA and 72/28 EVA. If it is assumed that the visual estimation method also gave a more reliable measure of the solubility of paclitaxel in polyurethane, that is, a value of greater than 30 % w/w, then a loading or total initial content (A) of 30 % w/w would have been less than the solubility. The release of paclitaxel from all films was biphasic (Figure 17). After the first several days the release of paclitaxel was not linear with the square root of time, rather, the release decreased more rapidly. This second phase of the biphasic release did not follow the Higuchi model because one or more of the assumptions were not valid. That is, the paclitaxel may not have been in a fine state such that the size of the particles were much less than the thickness of the films, but instead, during the casting of the films the paclitaxel may have aggregated as it precipitated out in the polymer matrix. The process of DCM evaporation from the films occurred relatively slowly, with probably most of the DCM evaporating under ambient conditions over approximately 1 h. During the evaporation of DCM from the films, the paclitaxel may have associated with one or more surfaces of the films. It is possible that as the DCM evaporated from the top side of the 117 films that the paclitaxel became concentrated towards the bottom side of the films and that the paclitaxel may not have been homogeneously distributed in the films. Surface associated paclitaxel may have been released from the films more rapidly than paclitaxel not associated with the surface of the matrices. Non-homogeneous distribution of drug in 60/40 EVA matrices has been overcome by Rhine (Rhine, 1980), who prepared protein drug loaded matrices in a manner that provided homogeneous drug distribution and reproducible release kinetics. These matrices were manufactured by dissolving the polymer and suspending the drug in DCM and then casting the mixture into molds that were pre-cooled to -80 °C. Most of the DCM was then evaporated from the matrices at -20 °C. Another assumption in the Higuchi model that may have possibly not been met, is that the amount of drug remaining in the matrices after several days may not have been much greater than the solubility of the paclitaxel in the polymer matrices. Rather, after the depletion of some of the drug, the remaining paclitaxel concentration in the polymer matrices may have been approximately equal to or less than the solubility of the drug in the matrices. Another assumption of the Higuchi release model is that sink conditions (taken to be a concentration of < 15 % of the solubility of the drug in the buffer) are maintained in the release buffer (Equation 34). For the films with the highest paclitaxel loading of 30 % w/w, the initial rapid release phase of paclitaxel from the various films resulted in non-sink conditions, for the first four to sixteen days of release. Though the films were sampled four times in the first day, these non-sink conditions existed due to concentrations of paclitaxel in the release media exceeding 15 % of the drug's solubility. The non-sink conditions over the first several days would not be expected to have 118 changed the cumulative amount of drug released, but would be expected to affect the rate of paclitaxel release from the 30 % w/w loaded films. 5.3. The effect of polymer type on the diffusion of paclitaxel through EVA and polyurethane films Plots of the cumulative amount of paclitaxel that had diffused through 60/40 EVA and polyurethane films (Figure 16) showed tL values of 1.95 and 2.41 d, respectively. The ti was due to the paclitaxel concentration in the films starting at zero. The paclitaxel diffused from solution in the donor cell, through the films, and into the receptor cell. After the period of tL, steady-state diffusion was reached. The values of h observed for 60/40 EVA and polyurethane films were 148 ± 6 and 135 ± 1 0 u.m, respectively. Values of D were calculated from ^ and h in Equation 43. The values of D for paclitaxel through 60/40 EVA and polyurethane films were calculated to be 2.16 x 10"10 and 1.46 x 10"10 cm2-s"', respectively. The slopes of the linear regression analysis lines (Figure 14) were used to calculate the values of P from Equation 41. The values of P for paclitaxel through 60/40 EVA and polyurethane films were calculated to be 3.85 x 10"5 and 2.67 x 10"5 cms"1, respectively. Values of K were calculated from h, P, and D in Equation 36. The values of K for paclitaxel through 60/40 EVA and polyurethane films were calculated to be 2 640 and 2 480, respectively. Values of CV were calculated from K and Q in Equation 39. The values of Cs- for paclitaxel in 60/40 EVA and polyurethane were 0.383 and 0.360 % w/v, respectively. 119 Maurin (1992) showed a ^  value of approximately 30 min for 4-chlorobenzoic acid through 60/40 EVA films with a thickness of 750 um. Although tL increases with an increase in h, it decreases with a decrease in the solubility of the drug in the donor cell buffer (Q). The value of Q for paclitaxel was 1.45 ug-mL"1, while the value of Cd for 4-chlorobenzoic acid was 110 ug-mL"1, approximately 75 times greater than that of paclitaxel. Maurin (1992) showed that the values of D for benzoic acids with various substituents, through 60/40 EVA, ranged from 0.45 x 108 to 2.26 x 108 c m V , while the values of P ranged from 3.22 x 108 to 72.7 x 108 cm2*"1. Maurin (1992) showed that the values K for benzoic acid with various substituents, through 60/40 EVA, ranged from 1.82 to 75.9. The values of K for paclitaxel were 33 to 1 450 times greater than the values of K for the various benzoic acids. The values of Cs- of benzoic acids with various substituents in 60/40 EVA, were calculated from the data from (Maurin, 1992) to range from 0.239 to 1.59 % w/v. The values of Cs-for paclitaxel were of approximately the same magnitude as the values of Cs- for the various benzoic acids. The difference in the values of K for paclitaxel and the various benzoic acids through 60/40 EVA are due to the differences in the values of Q for paclitaxel and the various benzoic acids, which were 1.45 jag-mL"1, and 110 to 5 330 ug-mL"1, respectively. The values of Cd for the various benzoic acids were 75 to 3 680 times greater than the value of Q for paclitaxel. The difference in the values of K between paclitaxel and the various benzoic acids partly accounts for the difference in the values of ti, D, and P between paclitaxel and the various benzoic acids. That is, partitioning of drug into the polymer is a major factor in limiting or regulating the diffusion of the drug through the polymer (Maurin, 120 1992). Paclitaxel has a greater MW (854 g-mol"1) than the various benzoic acids (152 to 157 g-mol"1), which also explains the greater values of ti, and lower values of D and P, for paclitaxel compared to the various benzoic acids. It will be more difficult for the larger molecule to diffuse through the polymer matrix than for the smaller molecules to do so. To summarize the diffusion parameters determined for paclitaxel in 60/40 EVA and polyurethane: The low solubility of paclitaxel in PBSA and its high solubility in EVA and polyurethane contributed to a large value of K for the drug. The large value of K and the large MW of paclitaxel contributed to low values of tL, D, and P for the drug. 5.4. The effect of monomer and polymer type on the release of paclitaxel from EVA and polyurethane films In vitro cumulative release profiles of paclitaxel plotted as a function of time, from films prepared from 60/40 EVA, 72/28 EVA, and polyurethane (Figure 15), show that the release was dependent on monomer and polymer type. The extent of paclitaxel release from 30 % w/w loaded 72/28 EVA films was less than that from 30 % w/w loaded 60/40 EVA films. According to the kinetics described for diffusion controlled release from a polymer matrix slab by Crank (Crank, 1975), when less than 50% of the loaded drug has been released, an effective diffusion coefficient (Peg) can be determined from Deff= ( M2n) I ( Cpo2t4 ) Equation 58 121 where Cpo is the original concentration of the drug in the polymer. The values of Deffwere calculated for 60/40 EVA, 72/28 EVA, and polyurethane at approximately day 3 and are summarized and compared to the values of D for 60/40 EVA and polyurethane in Table 11. Maurin (1992) showed that the permeability and partition coefficients of various benzoic acids with EVA increased nonlinearly with vinyl acetate content. This suggested to the authors that the diffusion of the benzoic acids across the EVA membranes was dependent on the partitioning of the acids into the polymer. The increase in partitioning of the benzoic acids with an increase in vinyl acetate content of the EVA was attributed to an increased interaction between the benzoic acids and the vinyl acetate portion of the polymer. The authors suggested that the partitioning of the benzoic acids into the EVA could be explained by a complex formation due to hydrogen bonding between the acidic hydrogen of the benzoic acids and the carbonyl portion of the vinyl acetate (Maurin, 1992). Given this complex formation put forward by Maurin (1992), governing of the partitioning, permeability, and diffusion of paclitaxel with 60/40 EVA, 72/28 EVA, and polyurethane, may be explained in part by a proposed complex formation between the drug and the polymers involving hydrogen bonds. This complexation is described in Figure 20. The hydroxyl groups of paclitaxel may hydrogen bond with the acetate groups of the EVA polymer. The acetate and the hydroxyl groups of paclitaxel may hydrogen bond with the urea groups of polyurethane. For a given polymer type, the larger the number of groups available for paclitaxel to complex with, the greater the value for Deg that would be expected. The values of Deff for 60/40 EVA, 72/28 EVA, and polyurethane, are given in Table 10. With a greater number of acetate groups available 122 for hydrogen bonding in 60/40 EVA, paclitaxel had a value of 2.86 x 10"11 cm -s" , while with a lesser number of acetate groups available for hydrogen bonding in 72/28 EVA, paclitaxel had a lower Devalue of 1.09 x 10"11 cm2-s"'. The difference in the values of diffusion for paclitaxel between EVA and polyurethane may be due in part to differences in the complexation between paclitaxel and the acetate groups of EVA, and the complexation between paclitaxel and the urea groups of polyurethane. In Table 11, the values of D, which were determined solely from drug diffusion studies, ranged from 1.46 x 10"10 to 2.16 x 10"10 cm2*"1, while the values ofDeff, determined solely from drug release studies, ranged from 2.86 x 10"" to 1.24 x 10"11 cm2-s"', for polyurethane and 60/40 EVA, respectively. As discussed in Section 5.2, non-sink conditions existed for the first four to sixteen days of the paclitaxel release studies. Sink conditions are an assumption in the determination of Deff, and non-sink conditions would be expected to result in lower values of D^than would be expected under sink conditions. The values of Z)^ were observed to be approximately an order of magnitude lower than the values of D. Table 11. Diffusion coefficient values, D^and D, of paclitaxel in 60/40 EVA, 72/28 EVA, and polyurethane films. Film type /VCcmV1) Db (cm2 s"1) 60/40 EVA 2.86 x l 0 " n ± 1.53 xlO"11 2.16 x 10"10 72/28 EVA 1.09 x 10"" ±0.63 x 10"u Polyurethane 1.24 x 10"11 +0.40 x 10"11 1.46 x 10"10 a Values of A;/r are the mean of 5, 10, and 30 % w/w loaded films ± standard deviation, determined at day 2.559. b Values of D are taken from Table 7. 123 Figure 20. Schematic of proposed complexation of paclitaxel with A) E V A and B) polyurethane. The hashed lines represent hydrogen bonds between the drug and the polymers. E V A - C H 2 — C H 2 — E V A O I C — C H 3 II O H I O I Paclitaxel B Polyurethane—O—C—N—Polyurethane O H,,,, II C—R I O I Paclitaxel Polyurethane— O—C—N—Polyurethane O R , 'O Paclitaxel Polyurethane—O—C—N—Polyurethane xvO H Paclitaxel 124 According to the kinetics described for diffusion controlled release from a saturated nondegradable matrix by Higuchi (1960) (Equation 44), the rate of drug release is indirectly proportional to the thickness of the film. The thickness of the various films ranged from 112 - 161 jam for polyurethane, 123 - 151 urn for 60/40 EVA, and 162 - 209 um for 72/28 EVA. Polymer type had an effect on film thickness in that 72/28 EVA films were thicker than 60/40 EVA and polyurethane films for some loadings of paclitaxel with or without sterilization. The greater thickness of some 72/28 EVA films compared to the other polymer types may be due to the lower density of 72/28 EVA (0.94 g-mL"1) compared to the density of 60/40 EVA (0.96 g-mL"1) (Salyer, 1971), and the density of polyurethane (1.05 g-mL"1) (Thermedics Inc. company literature, 1995). The 72/28 EVA films released less paclitaxel than the other polymer films, for some drug loadings. It may be that the thickness of the films had an influence on the release of drug. 5.5. The effect of drug loading on the release of paclitaxel from EVA and polyurethane films In vitro cumulative release profiles of paclitaxel plotted as a function of time, from films prepared from 60/40 EVA, 72/28 EVA, and polyurethane (Figure 17), showed that the release was dependent on drug loading. In vitro cumulative release of paclitaxel plotted as a function of the square root of time, from films prepared from 60/40 EVA, 72/28 EVA, and polyurethane (Figure 18), showed that the release was linear with the square root of time for the first several days. These observations are similar to those of Langer and his coworkers (Rhine, 1980) who demonstrated that an increase in protein drug loading increased the rate and extent of drug release from 60/40 EVA matrices, and 125 that the release of protein from these matrices was linear with the square root of time during the initial release of the drug. Miyazaki (Miyazaki, 1983) also observed that the release of 5-fluorouracil from poly(ethylene-co-vinyl alcohol) (EVOH) matrices was dependent on the initial drug loading, and that the cumulative release of drug from these matrices was linear with the square root of time during the initial release of the drug. 5.6. Selection of polymer for controlled release of paclitaxel for perivascular application for the inhibition of restenosis Paclitaxel loaded 60/40 EVA, 72/28 EVA, and polyurethane films all released the drug in vitro in a controlled manner. Advantages of controlled release formulations include a reduction in the total dose of drug required, a reduction in the frequency of administration, and a decrease in side effects (Thompson, 1960). Sterilization by y-irradiation, of 60/40 EVA, 72/28 EVA, and polyurethane films, resulted in cross-linking of the EVA matrices and chain scission of the polyurethane matrices. Sterilization by y-irradiation, of paclitaxel loaded 60/40 EVA, 72/28 EVA, and polyurethane films, also resulted in an increase in the elastic response of the EVA films, and an even greater increase in the elastic response of the drug loaded EVA films, while there was a small decrease in the elasticity of the polyurethane films. The polymer employed in the perivascular wrap should be biocompatible and sterile, which are requirements for implantation in vivo. Paclitaxel loaded polyurethane matrices appear to have been more stable to the effects of y-irradiation than the paclitaxel loaded EVA matrices. 126 All of the paclitaxel loaded 60/40 EVA, 72/28 EVA, and polyurethane films released the drug in vitro in a controlled manner and would be candidates for evaluation in animal models for the inhibition of restenosis. Given the effects of sterilization on the paclitaxel loaded 60/40 EVA, 72/28 EVA, and polyurethane films, the polyurethane films show the most promising potential for developing a film for the controlled release of paclitaxel for perivascular application for the inhibition of restenosis. 127 6. SUMMARY AND CONCLUSIONS 6.1. Physicochemical characterization of paclitaxel loaded 60/40 EVA, 72/28 EVA, and polyurethane films 1) The polymer crystallinity of paclitaxel loaded EVA films was low (2 to 4 %), whereas the polyurethane films were completely amorphous. 2) Paclitaxel was sparingly soluble in 60/40 EVA, whereas paclitaxel was freely soluble in polyurethane. 3) Paclitaxel existed within the EVA matrices as a granular, amorphous solid, whereas paclitaxel was miscible with the polyurethane matrices. 4) The polymer molecular weights of paclitaxel loaded, nonsterile 72/28 EVA, 60/40 EVA, and polyurethane films, were 23 300, 64 200, and 126 000 g-mol"1, respectively. 5) y-irradiation of paclitaxel loaded EVA films resulted in an increase in the elasticity of the films, whereas y-irradiation of polyurethane films resulted in a decrease in the elasticity of the films. 6) y-irradiation of paclitaxel loaded EVA films resulted in cross-linking of the polymer chains, whereas y-irradiation of polyurethane films resulted in polymer chain scission. 128 6.2. Paclitaxel in vitro diffusion through 60/40 EVA and polyurethane films 1) Partitioning, permeability, and diffusion coefficients of paclitaxel with EVA and polyurethane were determined to be similar for the two different types of matrices, and were explained in part by a proposed complex formation between the drug and the polymers involving hydrogen bonds. 6.3. Paclitaxel in vitro release from 60/40 EVA, 72/28 EVA, and polyurethane films 1) Paclitaxel release from EVA and polyurethane films was linear with both the square root of time, and the square root of the loading concentration, for the first several days. 2) Paclitaxel release from EVA and polyurethane films was by diffusion without the creation of channels or pores, and followed the Higuchi model of release for the first several days. 5) Paclitaxel release from 60/40 EVA, 72/28 EVA, and polyurethane, was influenced by polymer monomer ratio, polymer type, and drug loading. 129 6.4. Selection of polymer for paclitaxel loaded films for perivascular application for the inhibition of restenosis. 1) All of the paclitaxel loaded 60/40 EVA, 72/28 EVA, and polyurethane films released the drug in vitro in a controlled manner and would be candidates for evaluation in animal models for the inhibition of restenosis. 2) Given the effects of sterilization on paclitaxel loaded 60/40 EVA, 72/28 EVA, and polyurethane films, polyurethane films showed the most promising potential for developing a film for the controlled release of paclitaxel for perivascular application for the inhibition of restenosis. 130 7. FUTURE WORK There are at least three main areas for future work: 1) The proposed hydrogen bonding complexation of paclitaxel with EVA and polyurethane should be investigated further. The data that suggested paclitaxel partitioning, permeability, and diffusion were dependent on the amount of hydrogen bonding groups available in the EVA and polyurethane, should be confirmed. This work could be carried out by characterizing the diffusion of paclitaxel (with diffusion cells), with a series of EVA polymers with different ratios of ethylene to vinyl acetate, and with a series of polyurethanes with different ratios of urea to poly(tetramethylene ether glycol). 2) There may be an advantage in developing a biodegradable polymeric drug delivery system for paclitaxel that would release the drug in a controlled manner, and would completely biodegrade over time. 3) The release of paclitaxel from both sides of a film in vivo could potentially result in toxic effects of the paclitaxel on the healthy tissue surrounding the blood vessel in contact with the perivascular film. The development of a "unidirectional film" that would release paclitaxel only to the adventitia of the blood vessel, and not to the surrounding area, may overcome this potential drawback of the existing films. 131 APPENDIX I: SAMPLE CALCULATIONS OF DIFFUSION PARAMETERS Polyurethane is used in this example. 1) The thickness (h) of the films was determined as the mean ± standard deviation of the measured values of three films. h = 135 ± 1 0 urn 2) The cumulative amount of paclitaxel that had diffused through the polyurethane membrane (M) with time (f) was determined as the mean ± standard deviation of the measured values of three films. For example, at day six M= 25.9 ±5.2 jag 3) The graph (Figure 14) was prepared by plotting M with time (t) (Equation 41) where M = PSCc(t) - PSCc(t£) 4) The linear regression analysis line of the four terminal points was plotted and the graphing equation was found to be y = mx + b y = (7.09 ug-d-'Xx)-^.! u« 132 5) The theoretical time lag (tL) was determined by extrapolating the linear regression analysis line of the four terminal points to the x-axis. tL = 2.41 d 6) The diffusion coefficient (D) was calculated from Equation 43 and converted from units of [um2-d~'] to units of [cm2-s-1]. D = h2/6tL D = ((135 um) (1 cm/10 000 um))2 / ((6)(2.41 d)(86 400 s/1 d)) D= 1.46 x 1 0 " 1 0 c m V 7) The surface area of the membrane (S) was determined by measuring the area of the opening of the diffusion cells. 5=211 mm2 8) The solubility of the drug in the donor cell (CJ) was taken as the mean ± standard deviation of the saturated solubility values of three donor cells sampled daily over three days. Cd= 1.45 ±0.07 ug-mL"1 133 9) The permeability coefficient (P) was calculated from the slope of the linear regression analysis from Equation 41 where, using the format y = mx + b and M = PSCc(t) - PSCc{tL) thus P = m / SCd and converting units of [mm] and units of [mL] into units of [cm] gives P = ((7.09 -ug-d"1) / (((211 mm2)(l cm2 /100 mm2)) ((1.45 ug-mU'Xl mL /1 cm3)) P = 2.32 cm-d"1 10) The partition coefficient (K) (unitless) was calculated from Equation 36 K = PhlD and converting units of [d] into units of [s], and units of [urn] into units of [cm] ^=((2.32cm-d"1)(l d/86 40 s)(135 um)(l cm / 1000 um)) / (1.46 x lO - 1 0 cm2-s ^=2 480 ) The solubility (saturation concentration) of the drug in the polymer (CV) was calculated from Equation 39 Cs • — KCd and converting units of [|ag-mL" ] into units of [% w/v] Cs- = (2 480)(1.45 ug-ml/'XClOO % w/v) / (1 000 000 ug-mL"1) C 9' = 3.6%w/v 135 8. REFERENCES Abe, J., W. Zhou, J. Taguchi, N. Takuwa, K. Miki, H. Okazaki, K. Kurokawa, and M. Kumada. 1994. Suppression of neointimal smooth muscle cell accumulation in vivo by antisense CDC2 and CDK2 oligonucleotides in rat carotid artery. Biochemical and Biophysical Research Communications 198:16-24. Adams, I. and S.S. Davis. 1973. The formulation and sterilization of a surgical lubricant gel based on carboxypolymethylene. J.Pharm.Pharmac. 25:640-646. Ajani, J., D.H. Ilson, K. Daugherty, and D.P. Kelsen. 1995. Paclitaxel in the treatment of carcinoma of the esophagus. Seminars in Oncology 22(supp):34-40. Axel, D.L, W. Kunert, C. Goggelmann, M. Oberhoff, C. Herdeg, A. Kiittner, D.H. Wild, B.R. Brehm, R. Riessen, G. Koveker, and K.R. Karsch. 1997. Paclitaxel inhibits arterial smooth muscle cell proliferation and migration in vitro and in vivo using local drug delivery. Circulation 96:636-645. Banberger, R.L. 1989. Ethylene-vinyl acetate. In Modern plastics encyclopedia. Anonymous74-76. Barker, S.G.E., J.E. Beesley, P.A. Baskerville, and J.F. Martin. 1995. The influence of the adventitia on the presence of smooth muscle cells and macrophages in the arterial intima. Eur.J.Vasc.Endovasc.Surg. 9:222-227. 136 Barnes, H.A., J.F. Hutton, and K. Walters. 1989. Rheology Series, Vol. 3. An Introduction to Rheology. Anonymous Elsevier, New York. 1-10. Bawa, R., R.A. Siegel, B. Marasca, M. Karel, and R. Langer. 1985. An explanation for the controlled release of macromolecules from polymers. Journal of Controlled Release 1:256-267. Bennett, M.R., S. Anglin, J.R. McEwan, R. Jagoe, A.C. Newby, and G.I. Evan. 1994. Inhibition of vascular smooth muscle cell proliferation in vitro and in vivo by c-myc antisense oligodeoxynucleotides. J.Clin.Invest. 93:820-828. Bernabei, M.T., G. Gamberini, V. Ferioli, and R. Cameroni. 1982. Le forme cristalline del progesterone in elastomero siliconico. Bolletino Chimico Farmaceutico 121:347-353. Bernacca, G.M., M.J. Gulbransen, R. Wilkinson, and D.J. Wheatley. 1998. In vitro blood compatibility of surface-modified polyurethanes. Biomaterials 19:1151-1165. Besong, A.A., J.L. Tipper, E. Ingham, M.H. Stone, B.M. Wroblewski, and J. Fisher. 1998. Quantitative comparison of wear debris from UHMWPE that has and has not been sterilized by gamma irradiation. J.Bone Joint Surg. 80-B:340-344. 137 Chien, Y.W. 1978. Methods to achieve sustained drag delivery - the physical approach: implants. In Sustained and controlled release drug delivery systems. J.R. Robinson, editor. Marcel Dekker, Inc., New York. 252-253. Cleek, R.L., A.A. Rege, L.A. Denner, S.G. Eskin, and A.G. Mikos. 1997. Inhibition of smooth muscle cell growth in vitro by an antisense oligodeoxynucleotide released from poly(DL-lactic-co-glycolic acid) microparticles. J.Biomed.Mater.Res. 35:525-530. Clowes, A.W., M.A. Reidy, and M.M. Clowes. 1983b. Kinetics of cellular proliferation after arterial injury. Laboratory Investigation 49:327-333. Clowes, A.W., M.A. Reidy, and M.M. Clowes. 1983a. Mechanisms of stenosis after arterial injury. Laboratory Investigation 49:208-215. Cogswell, F.N. 1981. Polymer Melt Rheology. John Wiley & Sons, Toronto. 1-38. Cowie, J.M.G. 1973. Polymers: Chemistry and Physics of Modern Materials. Anonymous International Textbook Company Limited, Aylesbury. Crank, J. 1975. The mathematics of diffusion. Anonymous Clarendon Press, Oxford. 138 Dordunoo, S.K. and H.M. Burt. 1996. Solubility and Stability of Taxol: Effects of Buffers and Cyclodextrins. International Journal of Pharmaceutics 133:191-201. Dunn, R.L. 1991. Polymeric matrices. In Polymeric Drugs and Drug Delivery Systems. Anonymous 11 -23. Edelman, E.R., D.H. Adams, and M.J. Karnovsky. 1990. Effect of controlled adventitial heparin delivery on smooth muscle cell proliferation following endothelial injury. Medical Sciences 81:31133111-3111. Edelman, E.R., D.H. Adams, and M.J. Karnovsky, inventors. President and Fellows of Harvard College. 1996. Extraluminal regulation of the growth and repair of tubular structures in vivo. Mass/US. (5 527 532) A61K 9/12. 424/422; 424/423; 424/426; 424/430. Edelman, E.R., M.A. Nubent, L.T. Smith, and M.J. Karnovsky. 1992. Basic fibroblast growth factor enhances the coupling of intimal hyperplasia and proliferation of vasa vasorum in injured rat arteries. J.Clin.Invest. 89:465-473. Edelman, E.R., M. Simons, M.G. Sirois, and R.D. Rosenberg. 1995. c-myc in vasculoproliferative disease. Circ.Res. 76:176-182. 139 Faxon, D.P., W. Coats, and J. Currier. 1997. Remodeling of the coronary artery after vascular injury. Progress in Cardiovascular Diseases 40:129-140. Flynn, G.L., S.H. Yalkowsky, and T.J. Roseman. 1974. Review article: Mass transport phenomena and models: Theoretical concepts. Journal of Pharmaceutical Sciences 63:479-510. Francois, P., P. Vaudaux, N. Nurdin, H.J. Mathieu, P. Descouts, and D.P. Lew. 1996. Physical and biological effects of a surface coating procedure on polyurethane catheters. Biomaterials 17:667-678. Freese, A., B.A. Sabel, W.M. Saltzman, M.J. During, and R. Langer. 1989. Controlled release of dopamine from a polymeric brain implant: In Vitro characterization. Experimental Neurology 103:234-238. Golomb, G., I. Fishbein, S. Banai, D. Mishaly, D. Moscovitz, D. Gertz, A. Gazit, E. Poradosu, and A. Levitzki. 1996. Controlled delivery of a tyrphostin inhibits intimal hyperplasia in a rat carotid artery injury model. Atherosclerosis 125:171-182. Grulke, E.A. 1994. Polymer Process Engineering. Anonymous Prentice-Hall, Inc., Englewood Cliffs. 140 Hagan, S.A., S.S. Davis, L. Ilium, M.C. Davies, M.C. Garnett, D.C. Taylor, M.P. Irving, and Th.F. Tadros. 1995. Estimation of the poly(ethylene glycol) chain length of L-polylactide-polyethylene glycol in aqueous dispersions using viscoelastic measurements. Langmuir 11:1482-1485. Hall, C. 1981. Polymer materials. Anonymous The MacMillan Press Ltd., New York. Hearle, J.W.S. 1982. Polymers and their properties. Ellis Horwood Limited, Toronto. 27 pp. Heller, P.F., G.M. Jenkins, L. Cheng, K. Leong, W. Shao, J. Brinker, E. Hope, C. Nater, E. Lakatta, and J.P. Froehlich. 1995. The sustained release of Galardin and taxol from gelatin chondroitin sulfate coacervate films. Mat.Res. Soc. Symp.Proc. 394:55-60. Herembert, T., J. Gogusev, D.L. Zhu, T.B. Drueke, and P. Marche. 1997. Control of vascular smooth-muscle cell growth by macrophage-colony-stimulating factor. Biochem.J. 325:123-128. Higuchi, T. 1960. Rate of release of medicaments from ointment bases containing drugs in suspension. Journal of Pharmaceutical Sciences 50:874-875. 141 Higuchi, T. 1963. Mechanism of sustained-action medication. Theoretical analysis of rate of release of solid drugs dispersed in solid matrices. Journal of Pharmaceutical Sciences 52:1145-1149. Hosford, W.F. 1993. Tensile Testing. P. Han, editor. ASM International, Materials Park. 1-24. Huehns, T.Y., P. Gonschior, and B. Hofling. 1996. (Editorial) Adventitia as a target for intravascular local drug delivery. Heart 75:537-538. Huggins, M.L. 1942. The viscosity of dilute solutions of long-chain molecules. IV. Dependence on concentration. J.Am.Chem.Soc. 64:2716-2718. Joshi, R.R., T. Underwood, J.R. Frautschi, R.E. Phillips, F.J. Schoen, and R.J. Levy. 1996. Calcification of polyurethanes implanted subdermally in rats is enhanced by calciphylaxis. J.Biomed.Mater.Res. 31:201-207. Kakuta, T., M. Usui, W.D. Coats, J.W. Currier, Jr., F. Numano, and D.P. Faxon. 1998. Arterial remodeling at the reference site after angioplasty in the atherosclerotic rabbit model. Arterioscler. Thromb. Vasc.Biol. 18:47-51. Kelsen, D., J. Ajani, D. Ilson, K. Daugherty, and R. Pazdur. 1994. A phase II trial of paclitaxel (taxol) in advanced esophageal cancer: preliminary report. Seminars in Oncology 21:4422-48. 142 Kerns, E.H., K.J. Volk, and S.E. Hill. 1994. Profiling taxanes in Taxus extracts using LC/MS and LC/MS/MS techniques. J.Nat.Prod. 57:1391-1403. Kingston, D.G.I. 1991. The Chemistry of Taxol. Pharmac.Ther. 52:1-34. Kinsella, J.L. and S.J. Sollot, inventors. Sep 2, 1995. Method of treating atherosclerosis or restenosis using microtubule stabilizing agent. PCT (Patent Cooperation Treaty). (WO 95/13795) Kraemer, E.O. 1938. Molecular weights of celluloses and cellulose derivatives. Ind.Eng.Chem. 30:1200-1203. Kuntz, R.E. and D.S. Bairn. 1993. Defining coronary restenosis. Newer clinical and angiographic paradigms. Circulation 88:1310-1323. Lambert, T.L., V. Dev, E. Rechavia, J.S. Forrester, F. Litvack, and N.L. Eigler. 1994. Localized arterial wall drug delivery from a polymer-coated removable metallic stent. Circulation 90:1003-10011. Langer, R. andN. Peppas. 1983. Rev.Macromol.Chem.Phys. 61 Langer, R., H. Brem, and D. Tapper. 1981. Biocompatibility of polymeric delivery systems for macromolecules. J.Biomed.Mater.Res. 15:267-277. 143 Langer, R. and J. Folkman. 1976. Polymers for the sustained release of proteins and other macromolecules. Nature 263:797-799. Liggins, R.T., W.L. Hunter, and H.M. Burt. 1997. Solid-State Characterization of Paclitaxel. Journal of Pharmaceutical Sciences 86:1458-1463. Markman, M., E.K. Rowinsky, T. Hakes, B. Reichman, W. Jones, J.L. Lewis, and et al. 1992. Phase I trial of intraperitoneal taxol: a Gynecologic Oncology Group Study. JClin.Oncol. 10:1485-1491. Martin, A. 1993. Diffusion and Dissolution. In Physical Pharmacy. Anonymous324-361. Martini, L., J.H. Collett, and D. Attwood. 1997. The influence of gamma irradiation on the physicochemical properties of a novel triblock copolymer of 8-caprolactone and ethylene oxide. J.Pharm.Pharmacol. 49:601-605. Maurin, M.S., L.W. Dittert, and A.A. Hussain. 1992. Mechanism of diffusion of monosubstituted benzoic acids through ethylene-vinyl acetate copolymers. American Pharmaceutical Association 81:79-84. Maurin, N., N. Daty, C. Guernier, K. Dahmen, and H. Richter. 1997. An in vivo study of the biodegradation of the hydrophilic Mitrathane®. J.Biomed.Mater.Res. 34:73-78. 144 McLaughlin, J.L., R.W. Miller, R.G. Powell, and C.RJr. Smith. 1981. 19-Hydroxybaccatin III, 10-deacetylcephalomannine, and 10-deacetyltaxol: new antitumor taxanes from Taxus wallichiana. J.Nat.Prod. 44:312-319. Mills, S.N. and S.S. Davis. 1987. Controlled drug delivery. In Polymers in controlled drug delivery. L. Ilium and S.S. Davis, editors. IOP Publishing Limited, Bristol. 1-14. Mishaly, D., I. Fishbein, D. Moscovitz, and G. Golomb. 1997. Site-specific delivery of colchicine in rat carotid artery model of restenosis. Journal of Controlled Release 45:65-73. Miyazaki, S., Takeuchi, M. Sakamoto, and M. Takada, M. 1983. Controlled release of 5-fluorouracil from hydrophilic ethylene-vinyl alcohol copolymer matrices. Chemical and Pharmaceutical Bulletin. 31:3707-3713. Mohsen, N.M., R.G. Craig, and C T . Hanks. 1998. Cytotoxicity of urethane dimefhacrylate composites before and after aging and leaching. J.Biomed.Mater.Res. 39:252-260. Monsarrat, B., M. Wright, J. Dubois, and et al. 1990. Taxol metabolism: isolation and identification of three major metabolites of taxol in rat bile. Drug Metab.Dispos. 18:895-901. 145 Monsarrat, B., M. Wright, J. Dubois, and et al. 1992. Taxol metabolism in rat and human bile. Ann.Oncol. 3 (suppl 1): 123 Moore, G.R. and D.E. Kline. 1984. Properties and Processing of Polymers for Engineers. Anonymous Prentice-Hall, Inc., Englewood Cliffs. Morimoto, Y., T. Seki, K. Sugibayashi, K. Juni, and S. Miyazaki. 1988. Basic studies on controlled transdermal delivery of nicardipine hydrochloride using ethylene-vinyl acetate and ethylene-vinyl alcohol copolymer matrices. Chemical and Pharmaceutical Bulletin 36:2633-2641. Omelczuk, M.O. and J.W. McGinity. 1992. The influence of polymer glass transition temperature and molecular weight on drug release tablets containing poly(DL-lactic acid). Pharm.Res. 9:26-32. Parness, J., D.G.I. Kingston, R.G. Powell, C. Harracksingh, and S.B. Horwitz. 1982. Structure-Activity Study of Cytotoxicity and Microtubule Assembly In Vitro by Taxol and Related Taxanes. Biochemical and Biophysical Research Communications 105:1082-1089. Peng, T., P. Gibula, K. Yao, and M.F. A. Goosen. 1996. Role of polymers in improving the results of stenting in coronary arteries. Biomaterials 14:685-694. 146 Pitt, C.G. 1990. Poly-8-caprolactone and its copolymers. In Biodegradable polymers as drug delivery systems. M. Chasin and R. Langer, editors. Marcel Dekker Inc., New York. 71-120. Post, M.J., C. Borst, G. Pasterkamp, and C.C. Haudenschild. 1995. (Review) Arterial remodeling in atherosclerosis and restenosis: a vague concept of a distinct phenomenon. Atherosclerosis 118:S115-S123 Rhine, W.D., D.S.T. Hsieh, and R. Langer. 1980. Polymers for sustained macromolecule release: procedures to fabricate reproducible delviery systems and control release kinetics. Journal of Pharmaceutical Sciences 69:265-270. Ringel, I. and S.B. Horwitz. 1987. Taxol is converted to 7-epitaxol, a biologically active isomer, in cell culture medium. The Jounal of Pharmacology and Experimental Therapeutics 242:692-698. Rizzo, C , C. Riley, D. Von Hoff, J. Kuhn, T. Phillips, and T. Brown. 1990. Analysis of anticancer drugs in biological fluids: determination of taxol with application to clinical pharmacokinetics. Journal of Pharmaceutical and Biomedical Analysis 8:159-164. Rose, W.C. 1995. Preclinical antitumor activity of taxanes. In Taxol® science and applications. M. Suffhess, editor. 209-235. 147 Rosen, S.L. 1993. Fundamental Principles of Polymeric Materials. Anonymous John Wiley & Sons, Inc., Toronto. Rowinsky, E.K. and R.C. Donehower. 1995. Review article: Paclitaxel (Taxol). The New England Journal of Medicine 332:1004-1014. Rowinsky, E.K., N. Onetto, R.M. Canetta, and S.G. Arbuck. 1992. Taxol: the first of the taxanes, an important new class of antitumor agents. Semin.Oncology 19:646-662. Runt, J.P. 1985. Encyclopedia of Polymer Science and Engineering. Wiley, New York. 487 pp. Saltzman, W.M. and R. Langer. 1989. Transport rates of proteins in porous materials with known microgeometry. Biophysical Journal 55:163-171. Saltzman, W.M., S.H. Pasternak, and R. Langer. 1987. Quantitative image analysis for developing microstructural descriptions of heterogeneous materials. Chemical Engineering Science 42:189-204. Saltzman, W.M. and L.B. Tena. 1991. Spermicide permeation through biocompatible polymers. Contraception 43:497-505. 148 Salyer, 1.0. 1971. Structure and property relationships in ethylene-vinyl acetate copolymers. Journal of Polymer Science; PartA-1 9:3083-3103. Schiff, P.B., J. Fant, and S.B. Horwitz. 1979. Nature 277:665-667. Schiff, P.B. and S.B. Horwitz. 1980. Proc.Natn.Adad. 77:1561-1565. Schwartz, S.M., C E . Murry, and E.R. O'Brien. 1996. Vessel wall response to injury. Science & Medicine 3:12-21. Sellke, F.W., J. Li, A. Stamler, J.J. Lopez, K.A. Thomas, and M. Simons. 1996. Angiogenesis induced by acidic fibroblast growth factor as an alternative method of revascularization for choronic myocardial ischemia. Surgery 120:182-188. Sen, M. and O. Gtiven. 1995. A comparative study of thermal and mechanical stabilities of gamma irradiated ethylene - ethyl acrylate and ethylene - vinyl acetate copolymers. Radiat.Phys.Chem. 46:871-874. Shah, V.P., K.K. Midha, S. Dighe, I.J. McGilveray, J.P. Skelly, A. Yacobi, T. Layloff, C T . Viswanathan, C E . Cook, R.D. McDowall, K.A. Pittman, and S. Spector. 1992. Analytical methods validation: Bioavailability, bioequivalence and pharmacokinetic studies. Pharmaceutical Research 9:588-592. 149 Sharma, M.G. 1965. Testing of Polymers. Vol. I. J.V. Schmitz, editor. Interscience Publishers, New York. 147-199. Siegel, R.A., J. Kost, and R. Langer. 1989. Mechanistic studies of macromolecular drug release from macroporous polymers. I. Experiments and preliminary theory concerning completeness of drug release. Journal of Controlled Release 8:223-236. Siegel, R.A. and R. Langer. 1990. Mechanistic studies of macromolecular drug release from macroporous polymers. II. Models for the slow kinetics of drug release. Journal of Controlled Release 14:153-167. Sintzel, M.B., A. Merkli, C. Tabatabay, and R. Gurny. 1997. Influence of irradiation sterilization on polymers used as drug carriers - a review. Drug Development and Industrial Pharmacy 23:857-878. Skarja, G.A. and J.L. Brash. 1997. Physicochemical properties and platelet interactions of segmented polyurethanes containing sulfonate groups in the hard segment. J.Biomed.Mater.Res. 34:439-455. Sollot, S.J., M.T. Cheng, R.R. Pauly, G.M. Jenkins, R.E. Monticone, M. Kuzuya, J.P. Froehlich, M.T. Crow, E.G. Lakatta, E.K. Rowinsky, and J.L. Kinsella. 1995. Taxol inhibits neointimal smooth muscle cell accumulation after angioplasty in the rat. Journal of Clinical Investigation 95:1869-1876. 150 Solomon, O.F. and I.Z. Ciuta. 1962. Determination de la viscosite intrinseque de solutions de polymers par une simple determination de la viscosite. J.Appl.poly.Sci. VI-24:683-686. Song, D., L.-F. Hsu, and J.L.S. Au. 1996. Binding of Taxol to plastic and glass containers and protein under in vitro conditions. Journal of Pharmaceutical Sciences 85:29-31. Tamargo, R.J., K.W. Leong, and H. Brem. 1990. Growth inhibition of the 9L glioma using polymers to release heparin and cortisone acetate. Journal ofNeuro-Oncology 9:131-138. Thermedics Inc. company literature. 1995. Medical grade aliphatic thermoplastic polyurethanes. pp. 1-8. Thompson, R.E. 1960. Bull.Parent.Drug 14:6-17. Tobolsky, A.V.e. and H.F.e. Mark. 1971. Polymer Science and Materials. A.V. Tobolsky and H.F. Mark, editors. John Wiley and Sons, Inc., Toronto. Turner, S. 1993. Tensile Testing. P. Han, editor. ASM International, Materials Park. 105-133. 151 Vandelli, M.A., F. Forni, and R. Cameroni. 1993. Analysis of release data in the evaluation of the physical state of progesterone in matrix systems. Journal of Microencapsulation 10:55-95. Volland, C , M. Wolff, and T. Kissel. 1994. The influence of terminal gamma-sterilization on captopril containing poly(D,L-lactide-co-glycolide) microspheres. Journal of Controlled Release 31:293-305. Wani, M . C , H.L. Taylor, M.E. Wall, P. Coggon, and A.T. McPhail. 1971. Plant antitumor agents. VI. The isolation and structure of taxol, a novel antileukemic and antitumor agent from Taxus brevifolia. Journal of the American Chemical Society 93:2325-2327. Wise, E.L., G.J. McCormick, G.P. Willet, L.C. Anderson, and J.F. Howes. 1978. J.Pharm.Pharmac. 30:686-689. Zutty, N.L. and R.D. Burkhart. 1964. Quantitative aspects of high pressure copolymerizations involving ethylene as a principal component. In Copolymerization. G.E. Ham, editor. Interscience Publishers, New York. 639-652. Zar, J.H. 1984. Biostatistical Analysis. 2nd Ed. Prentice Hall, Inc. Englewood Cliffs, N.J. 31-577. 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.831.1-0089048/manifest

Comment

Related Items