OPHTHALMIC DRUG DELIVERY BY SOFT CONTACT LENSES

Transcription

1 OPHTHALMIC DRUG DELIVERY BY SOFT CONTACT LENSES By JINAH KIM A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA

2 2008 Jinah Kim 2

3 To my loving husband Byoung Sam and my precious daughter Dahyoung. 3

4 ACKNOWLEDGMENTS I thank my advisor and supervisory committee chair, Dr. Anuj Chauhan, for his continuous guidance and encouragement on my research as well as great understandings and support on my life. He has been willing to give me lots of heartwarming advice professionally and also personally. He is the best advisor that I have ever had. I also thank my committee members, Dr. Spyros A. Svoronos, Dr. Kirk J. Ziegler, and Dr. Gregory S. Schultz for their time and support for this work. I would like to thank all our group members: Yash Kapoor who yielded lots of time and research subjects to me all the time, Brett Howell, Chhavi Gupta, Hyun Jung Jung, Chen-Chun Peng, and former members, Dr. Chi-Chung Li, Dr. Heng Zhu, Dr. Chen Zhi, and Dr. Marissa Fallon. Their friendship, help, and encouragement are unforgettable. I also thank the undergraduate students: Anthony B. Conway helped me with synthesizing transparent silicone hydrogel lenses; and Joshua R. McCarty and Andy Cohen with different experiments. In addition, I would like to gratefully acknowledge the technical support for this research by James Hinnant and Dennis Vince. I also thank all the other staffs in the department of Chemical Engineering at University of Florida. The help with characterizing hydrogels and particles by the people at Particle Engineering Research Center (PERC) and at Major Analytical Instrument Center (MAIC) at UF is appreciated. I thank my parents and my brother who are always with me for their unchanging love and support with pray throughout my life. Their enormous love that they have shown to me is not describable. I would like to give additional thanks to my other family members for their love. Most of all, I thank my dear husband, Byoung Sam, who totally supports me to pursue PhD. His love and understandings as well as his scarification of his lots opportunities in his career for the 4

5 support are invaluable. I am also indebted to my precious daughter, Dahyoung. I am thankful for her existence. Finally, I am grateful to God (who is always with me) for His sincere love and for guarding me. 5

6 TABLE OF CONTENTS ACKNOWLEDGMENTS...4 LIST OF TABLES...9 LIST OF FIGURES...10 LIST OF ABBREVIATIONS...14 ABSTRACT...15 CHAPTER 1 INTRODUCTION DEXAMETHASONE TRANSPORT AND OCULAR DELIVERY FROM POLY(HYDROXYETHYL METHACRYLATE) GELS...22 page 2.1 Introduction Materials and Methods Materials Synthesis of Poly(Hydroxyethyl Methacrylate) (PHEMA) Gels Drug Loading Drug Release Experiments Effect of Gel Thickness on Loading and Release of Drug Effect of Ionic Strength in Release Solution on Loading and Release of DXP Conversion of UV-VIS Absorbance to the Corresponding Concentration of Drug Determinations of Critical Micelle Concentration (CMC) of DXP Results and Discussion Dexamethasone (DX) Loaded Gel Loading (by soaking in aqueous solution) and release studies of DX Release of DX loaded in the gel by direct entrapment Effect of gel thickness on DX loading Effect of crosslink density of gel on DX loading Release of DX into PBS Dexamethasone 21-acetate (DXA) Loaded Gel Release of DXA loaded by direct entrapment Release of DXA loaded by DXA-ethanol presoaking Dexamethasone 21-disodium phosphate (DXP) Loaded Gel Partition coefficient and dynamics of DXP loading and release Effect of gel thickness on DXP loading Effect of ionic strength of outer solution on DXP loading and release

7 2.3.4 Mathematical Model for Drug Transport to the Cornea Concentration profiles in the POLTF Fraction of drug that enters cornea Conclusions EXTENDED DELIVERY OF OPHTHALMIC DRUGS BY SILICONE HYDROGEL CONTACT LENSES Introduction Materials and Methods Materials Preparation of Silicone Hydrogels Drug Loading Drug Release Experiments Packaging Tests Gel Characterization Dynamic mechanical analysis Ion permeability measurements Surface contact angle measurements Transmittance measurements Results and Discussion Timolol Loaded Gel Loading and release of timolol by the GELS 1 and 4 soaked in PBS solution Release of timolol from GELS 1-6 after loading by soaking in drugethanol solution Rate limiting mechanism for drug transport Impact of concentration in drug-ethanol solution on drug loading and release DX or DXA Loaded Gel Loading and release of DX by the gel soaked in PBS solution Release of DX(or DXA) loaded in the gel by soaking in drug-ethanol solution Packaging Effects Mechanical Properties of Gels Ion Permeability, Equilibrium Water Content, and Surface Contact Angle of Gels Transparency of Gels Conclusions EXTENDED DRUG DELIVERY FROM SILICONE-HYDROGEL CONTACT LENSES CONTAINING DIFFUSION BARRIERS Introduction

8 4.2 Materials and Methods Materials Drug Loading into Pure Lenses Vitamin E Loading into Pure Lenses Drug Loading into Vitamin E Loaded Lenses Drug Release Experiments Results and Discussion Dynamics of Drug By Pure Contact Lenses DX loaded lenses Timolol loaded lenses Effect of Vitamin E Loading in Water Content of Lenses Effect of Vitamin E Loading in Size of Contact Lenses Effect of Vitamin E Loading on Dynamic of Drug DX-vitamin E loaded lenses Timolol-vitamin E loaded lenses Scaling theory Diffusivities of DX or timolol in vitamin E loaded lenses Conclusions CONCLUSIONS FUTURE WORK Drug Transport in PHEMA Copolymer Silicone Contact Lenses for Extend Delivery of Ophthalmic Drug Drug Transport in Commercial Silicone Contact Lenses APPENDIX MODEL DESCRIPTION FOR DRUG DELIVERY BY SOAKED CONTACT LENSES LIST OF REFERENCES BIOGRAPHICAL SKETCH

9 LIST OF TABLES Table page 2-1 Concentration and partition coefficients (K) of DX in a soaking method Apparent partition coefficients (K app ), intrinsic partition coefficients (K ), and permanently entrapped drug amount (M p ) for DX release from the PHEMA gel synthesized in a direct entrapment method Effect of crosslink density on partition coefficient and diffusivities of DX in the gel synthesized in a direct drug entrapping method Apparent partition coefficients (K app ), intrinsic partition coefficients (K ), and permanently entrapped drug amount (M p ) for DXA release from the PHEMA gel synthesized in a direct entrapment method Concentration and partition coefficients (K) of DXP in a soaking method Effect of ionic strength (I) on DXP loading and release Effect of ionic strength on timolol maleate loading and release Various fractions (F c, F s, and F p ) in the eye for three derivatives of dexamethasone Compositions of the monomer mixtures (in ml) for various gels (0.18 ml of NVP and 15 µl of EGDMA was added to each composition before preparing the gels) Drug concentration and partition coefficients (K) of drug Diffusivities of timolol in gels Physical properties of gels List of silicone containing commercial contact lens (dipoter -6.50) used in this study. (n=6) Changes in diameter and water content of vitamin E loaded contact lenses before and after loading vitamin E Products of partition coefficient of DX and lens volume (KV lens ) for lenses soaked in DX-PBS solution Parameters of scaling theory for timolol Ratio of diffusivities of drug in vitamin E loaded lens to pure lens (D/D 0 ) A-1 Model parameters for three derivatives of dexamethasone

10 LIST OF FIGURES Figure page 2-1 Molecular structures of model drugs A) DX B) DXA C) DXP The model geometry of the gel Comparison of the model prediction and experimental data for DX loading into PHEMA gel soaked in drug solution Comparison of the model prediction and experimental data for DX release into fresh water from a PHEMA gels which have been soaked in drug solution Comparison of the model prediction and experimental data for DX release into the fresh water from PHEMA gels synthesized in a direct entrapment method Effect of gel thickness on DX loading and release of a PHEMA gel soaked in drug solution Comparison of DX release profiles in water and PBS from a PHEMA gel synthesized in a direct entrapment method Comparison of the model prediction and experimental data for DXA release into the fresh water from PHEMA gels synthesized in a direct entrapment method Effect of gel thickness on DXA release of a PHEMA gel which has been soaked in DXA-ethanol solution Comparison of DXA release profiles in water and PBS from a PHEMA which has been soaked in DXA-ethanol solution Comparison of the model prediction and experimental data for DXP loading by PHEMA gel soaked in DXP solution Comparison of the model prediction and experimental data for DXP release into the PBS from a PHEMA gels which have been soaked in drug solution Effect of gel thickness on DXP loading to a PHEMA soaked in DXP solution Effect of ionic strength on DXP loading and release in a soaking method Plot of diffusivities of DXP for loading and release in a soaking method as a function of ionic strength Effect of ionic strength on timolol maleate loading and release in a soaking method Plot of diffusivities of timolol maleate for loading and release in a soaking method as a function of ionic strength. C w,i = mg/ml

11 2-18 Plot of surface tension of DXP as a function of concentration for determination of CMC Concentration transients of DX in the POLTF at different axial locations for case 1, i.e., no flux to the PLTF Profiles of timolol uptake and release from 0.1 mm thick A) GEL1 B) GEL4 in PBS solution Profiles of timolol release from 0.1 mm thick GELS Profiles of timolol release from 0.2 mm thick A) GEL7 B) GEL Continued Profiles of timolol release from three different gels (0.4 mm thick) vs. scaled time Plot of mass of timolol released verses square root of time Profiles of DX uptake by 0.1 mm thick A) GEL1 B) GEL4 in PBS solution Profiles of DX release from four different composition gels (0.1 mm thick). DX was loaded in the gel by soaking in drug-ethanol solution of 4.99 mg/ml Effect of gel thickness on DX release from (a) GEL1 (b) GEL Plot of mass of DX released verses square root of time (0.1mm thick gels) Profile of DXA release from a 0.4 mm thick GEL Profiles of timolol release from the GEL1 packaged for 1.3 and 2 month Profiles of DXA release from the GEL1 packaged for 1.5 and 2 month Dependence of storage moduli of the gels on frequency Dependence of loss moduli of the gels on frequency Effect of DX loading method on profile of DX release by A) ACUVUE ADVANCE B) ACUVUE OASYS C) NIGHT&DAY D) O 2 OPTIX E) PureVision contact lenses. F) The plot of (DX release time) -1 versus water content of contact lenses Continued Continued

12 4-2 Effect of timolol loading method on profile of timolol release by A) ACUVUE ADVANCE B) ACUVUE OASYS C) NIGHT&DAY D) O 2 OPTIX E) PureVision contact lenses Continued Continued Correlation of vitamin E loading and concentration of soaking solution for different lenses Plot of A) water content (Q) B) EW of vitamin E loaded lenses versus vitamin E loading Percent increase in diameter of A) dry lenses B) wet lenses before and after loading vitamin E. Lines are best fit straight lines passing zero to the data Profiles of DX uptake and release by vitamin E loaded contact lenses A) ACUVUE OASYS B) NIGHT&DAY C) O 2 OPTIX D) PureVision. Solid legends represent uptake and the hollow legends release Continued Profiles of timolol release by vitamin E loaded contact lenses. Timolol and vitamin E were loaded together by soaking A) ACUVUE OASYS B) NIGHT&DAY C) O 2 OPTIX D) PureVision contact lens in timolol/vitamin E-ethanol solution (0.8 mg timolol in 1 ml of vitamin E-ethanol solution of various concentrations) for 24 hours Continued Profiles of timolol release by vitamin E loaded contact lenses A) NIGHT&DAY B) O 2 OPTIX Duration increase of A) DX loading B) DX release C) and D) timolol release by vitamin E loaded contact lenses Continued Profile of timolol release by TRIS loaded NIGHT&DAY lens Plot of % drug release by vitamin E loaded lenses versus square root of time Slope of best fit straight line to the plot of A) % DX release B) % timolol release versus square root of time for short time by vitamin E loaded lenses (Figure 4-11) Plot of D 0 /D versus vitamin E loading for timolol release

13 4-14. Plot of % drug release by vitamin loaded lenses versus (Dh 0 2 /D 0 h 2 )t. A) ACUVUE OASYS B) NIGHT&DAY contact lens A-1 Geometry of lens utilized in the model

14 LIST OF ABBREVIATIONS CMC DI DMA DX DXA DXP EGDMA EW FEP HEMA NVP PBS PHEMA PLTF POLTF TRIS Critical micelle concentration Deionized N,N-dimethylacrylamide Dexamethasone Dexamethasone 21-actate Dexamethasone 21-disodium phosphate Ethylene glycol dimethacrylate Equilibrium water content A polymer of tetrafluoroethylene and hexafloropropylene 2-Hydroxyethyl methacrylate 1-Vinyl-2-pyrrolidone Dulbecco s phosphate buffered saline Poly(hydroxyethyl methacrylate) Pre-lens tear film Post-lens tear film 3-Methacryloxypropyltris(trimethylsiloxy)silane 14

15 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy OPHTHALMIC DRUG DELIVERY BY SOFT CONTACT LENSES Chair: Anuj Chauhan Major: Chemical Engineering By Jinah Kim May 2008 Drug delivery via contact lenses is known to significantly increase the bioavailability of ophthalmic drugs compared to that via eye drops. Our study focused on investigating the feasibility of delivery ophthalmic drug for extended period of time by contact lenses. To have the fundamental insight of drug transport in hydrogel contact lenses, we explored ocular delivery of dexamethasone via poly(hydroxylethyl methacrylate) contact lenses, which are common daily disposable contact lenses. Three derivatives of dexamethasone (dexamethasone 21- disodium phosphate (DXP), dexamethasone (DX), and dexamethasone 21-acetate (DXA)) are explored. These drugs are loaded in the gels by soaking in aqueous or ethanol solutions, and also by direct addition of the drug to the polymerizing mixture. Dynamic drug concentrations in the aqueous phase are monitored both in loading and release experiments. The data is utilized to determine the partition coefficients and the mean diffusivity, which includes contributions from both bulk and surface diffusion. Finally we utilize the transport model to predict the bioavailability of the three forms of dexamethasone for drug delivery via contact lenses. While hydrogel contact lenses lead to higher bioavailability, these cannot be used for extended drug delivery because these cannot be worn overnight. Silicone hydrogel contact lenses can be worn for extended periods but commercial silicone hydrogel lenses are not suitable for 15

16 extended drug delivery. So we developed new extended wear silicone hydrogel soft contact lenses that deliver ophthalmic drugs for an extended period of time ranging from weeks to months. Silicone hydrogels comprising of N,N-dimethylacrylamide, 3-methacryloxypropyltris (trimethylsiloxy) silane, bis-alpha,omega-(methacryloxypropyl) polydimethylsiloxane, 1-vinyl- 2-pyrrolidone, and ethylene glycol dimethacrylate were prepared with varying ratios of monomers and transport of three different ophthalmic drugs, timolol, dexamethasone, and dexamethasone 21-acetate was explored. All the silicone hydrogels of 0.1 mm thickness exhibit diffusion limited transport and extended release varying 20 days up to more than three months depending on the compositions of hydrophobic and hydrophilic components of silicone hydrogels. Also, there are multiple time scales in transport of at least certain molecules, which is perhaps due to the complex microstructure of these gels. The mechanical and physical properties of lenses such as ion permeability, equilibrium water content, transparency, and surface contact angles of some of the gels are suitable for contact lens application. We also focused on the concept of increasing the release duration of drugs from commercial silicone contact lenses by creation of transport barriers. As diffusion barriers, we loaded the vitamin E, which is considered as an important ocular nutraceutical, into five different commercially available silicone contact lenses and transport of dexamethasone and timolol were measured. The results conclusively show that vitamin E loading can substantially increase the release duration of drugs, particularly for hydrophilic drugs. The effect of vitamin E loading on the release rates is qualitatively similar for ACUVUE OASYS, NIGHT&DAY, and O 2 OPTIX. The effect of vitamin E loading in dynamics of both DX and timolol is minimal for PureVision. 16

17 CHAPTER 1 INTRODUCTION In the last few decades, a number of novel approaches have been developed for controlled drug delivery of ophthalmic drugs. However, topical application of eye drops is still the dominant treatment methodology despite its inefficiency [1]. After application of an eye drop, the drug solution mixes with tear fluid, and then within about 5 minutes, a majority of drug is eliminated by tear drainage and conjunctival uptake. Due to the short residence time, only about 1-5% of the applied drug penetrates the cornea and reaches the intraocular tissues. To maintain therapeutic levels of drug concentration, frequent instillation of drops with large drug loadings are required, which is inconvenient for patients. Moreover, the drug that gets absorbed in conjunctiva or in the nasal cavity eventually reaches other organs through the blood circulation leading to side effects [2, 3]. To achieve increase ocular bioavailability, researchers have explored a variety of vehicles including suspension of nanoparticles, nanocapsules, liposomes and niosomes, ocular inserts like collagen shields and Ocusert, and therapeutic contact lenses. Among theses, contact lenses have been widely studied due to the high degree of comfort and biocompatibility. On instillation of medicated contact lens in the eye, drug diffuses through the lens matrix into the thin tear film named post-lens tear film (POLTF) trapped between the lens and the cornea, and the drug has a residence time about 30 min in the eye [4, 5]. An increase in the residence time leads to a significant increase in the bioavailability. Both mathematical models and clinical data suggest that the bioavailability for ophthalmic drug delivery can be as large as 50%, which is an order of magnitude larger than that for drops [6]. There have been a number of attempts in the past to use contact lenses for ophthalmic drug delivery; however most of these focused on soaking hydrophilic lenses in a drug solution followed by insertion into the eye [7 13]. The major problem of loading drug by this method is 17

18 that in most cases the loading capacity of the soaked contact lenses is inadequate. Another commonly used method of entrapping drugs in gels is direct addition of drug in the polymerizing medium [14 16]. Although such direct loading of drug into the lenses can allow higher loadings of the drugs, it can result in an activity loss during polymerization. Furthermore, a majority of the drug can diffuse from the lenses into the packaging medium and the drug retained in the lens can diffuse from the lens rapidly after insertion into the eye. In chapter 2, we explored ocular delivery of dexamethasone via poly(hydroxylethyl methacrylate) contact lenses to understand the drug transport by the contact lens in the fundamental point of view. Dexamethasone is a glucocorticoid steroid, which is similar to the natural steroid hormone made by the adrenal glands in the body. It relieves eye inflammation and swelling, heat, redness, and pain caused by chemicals, infection, and/or severe allergies. It is also used to treat persistent macular oedema in retina, which is a major cause of visual disabilities and blindness among individuals with diabetes [17]. Prolonged systemic administration of steroid can cause serious side effects such as diabetes, hemorrhagic ulcers, skin atrophy, myopathies, osteoporosis and psychosis [18]. Furthermore, it has been reported that continuous application of eye drops of 0.1% dexamethasone for extended periods of time (varying from 3 weeks to a year) can cause glaucoma accompanied by optic nerve damage, defects in visual acuity and fields of vision, and posterior subcapsular cataract formation and thinning of the cornea or sclera [19, 20]. Controlled drug delivery of dexamethasone to the eye through daily wear contact lens is expected to be safer than delivery via drops because of reduction in the amount of drug that reaches other body tissues through systemic circulation [6]. Several forms of dexamethasone with significantly different aqueous solubilities have been utilized in ocular studies including dexamethasone 21- disodium phosphate, dexamethasone, and dexamethasone 21-acetate 18

19 (arranged in the order of increasing partition coefficients measured in octanol-phosphate buffer solution) [21, 22]. Molecular structures of these three dexamethasone derivatives are shown in Figure 2-1. Dexamethasone and dexamethasone 21- acetate are hydrophobic drug while dexamethasone 21- disodium phosphate is ionic and thus freely water soluble. Civiale et al. screened the ocular permeability of dexamethasone derivatives through cell culture (in vitro) and excised rabbit cornea (ex vivo) and studied in vivo concentration of dexamethasone in rabbit aqueous humor as well. They reported that the permeability rates of dexamethasone derivates through cornea generally increase as octanol/water partitioning coefficients (log P) increase [22]. Weijtens et al. determined the dexamethasone concentration in aqueous humor, vitreous and serum after dexamethasone administration through various routes such as topical application of eye drop, subconjunctival injection, a peribulbar injection, and an oral dose [23 27]. They showed that the dexamethasone concentration in the aqueous humor is far lower for eye drops of 0.1% dexamethasone disodium phosphate compared to a subconjunctival injection even if an eye drop is instilled every 1.5 hour. However the conjunctival injection is also not the optimal drug delivery vehicle because it needs to be applied daily to have sufficiently high dexamethasone concentrations in the aqueous humor. It is thus desirable to have other drug delivery vehicles that can deliver dexamethasone in a non invasive manner, and yet achieve sufficiently high concentrations in aqueous humor. To address the issues of insufficient drug loading by soaking and rapid diffusion of drug by direct loading in the polymerizing medium Gulsen and Chauhan have proposed the development of nanoparticle laden gels that can load substantial amount of drug to the gel which can be released at a controlled rate from the nanoparticles [28, 29]. Also, a number of researchers have focused on developing biomimetic and imprinted contact lenses [30 35]. It has been 19

20 demonstrated both by in vivo and in vitro studies that soft contact lenses fabricated by the molecular imprinting method have a drug loading capacity 2- to 3-fold greater than that of the contact lenses made by a conventional method and leads to an increase in the partition coefficients and slower release of drugs. While the approaches listed above are effective at increasing the release duration from contact lenses, the lenses are still not suitable for extended release lasting a week or longer. Additionally, the studies cited above focused on hydrophilic hydrogel based contact lenses, which are not suitable for extended wear due to limited oxygen permeability. Due to high oxygen permeability silicone hydrogel contact lenses can be prescribed for extended wear lasting several weeks, and so these lenses are most suitable for development of extended drug delivery vehicles. Recently Karlgard et. al. measured the uptake and release of a number of ophthalmic drugs by commercially available HEMA based and silicone contact lenses in vitro studies [36]. A number of drugs including cromolyn sodium, ketotifen fumarate, ketorolac tromethamine, and dexamethasone sodium phosphate which are all hydrophilic were loaded into these lenses by soaking these in drug solutions for a limited period of time. The release studies showed that a majority of the drug taken up by the gels was released in a short period of time, proving that commercial silicone hydrogel contact lenses do not have the appropriate composition to provide extended drug release. Also they showed that the type of drug and the type of material affect the drug uptake and release In the chapter 3, we developed new extended wear silicone hydrogel soft contact lenses that deliver ophthalmic drugs for an extended period of time ranging from weeks to months. The important properties for contact lens applications were explored and furthermore we focused on 20

21 correlation between composition and transport mechanism of drug such as timolol and dexamethasone in the silicone hydrogels. A number of approaches have been explored in the past to increase the duration of drug release from contact lenses. These include dispersing colloidal particles in the gels, molecular imprinting and biomimetic recognition. All these approaches focus on increasing the affinity of the gel for the drug. In chapter 4, we propose a novel approach of creating diffusion barriers in the contact lenses by loading the lenses with excipients that have a very small affinity for the drug. If a drug has negligible affinity for the excipients, drug molecules must diffuse around the regions containing the excipients leading to longer diffusive path lengths, which must reduce the mean diffusivity. Essentially, the proposed role of the excipients is to increase the tortuosity of the gel for drug transport. It is important to choose excipients with a very small aqueous solubility so that it does not diffuse into the tear film in large amounts. It is also important to choose an excipient that does not have ant toxic effects on the cornea. We focus on using vitamin E as the excipient because of very low solubility of hydrophilic drugs in vitamin E, and also its high viscosity, low aqueous solubility and a high degree of compatibility with ocular tissue. 21

22 CHAPTER 2 DEXAMETHASONE TRANSPORT AND OCULAR DELIVERY FROM POLY(HYDROXYETHYL METHACRYLATE) GELS 2.1 Introduction The aim of this chapter was to investigate loading and release of different forms of dexamethasone in PHEMA gels, which are a common contact lens material. Three different derivatives of dexamethasone, i.e., dexamethasone (DX), dexamethasone 21-actate (DXA), and dexamethasone 21-disodium phosphate (DXP) were incorporated in the gel by soaking gels in drug solutions (soaking method) or by direct dissolution of the drug in the polymerizing mixture (direct entrapment method). The loaded drug was then released by soaking the drug-loaded gels in DI water or PBS. Dynamic drug concentrations in the aqueous phase were monitored both in loading and release experiments. The equilibrium uptake in these experiments was utilized to determine the partition coefficients, and the dynamic data was fitted to a modified diffusion equation to determine the mean diffusivity, which includes contributions from both bulk and surface diffusion. Finally the transport model for the drugs was utilized to predict the bioavailability of the three forms of dexamethasone for drug delivery via PHEMA contact lenses Materials 2.2 Materials and Methods 2-hydroxyethyl methacrylate (HEMA, 97%) monomer, dexamethasone 21-acetate (DXA, 99%), dexamethasone 21-disodium phosphate (DXP, 99%), timolol maleate, ( 98%), ethanol ( 99.5%), and Dulbecco s phosphate buffered saline (PBS) were purchased from Sigma-Aldrich Chemicals (St. Louis, MO); dexamethasone (DX, 98%) and ethylene glycol dimethacrylate (EGDMA) from Sigma-Aldrich Chemicals (Milwaukee, WI). Darocur TPO was kindly provided by Ciba Specialty Chemicals (Tarrytown, NY). Nitrogen was bought from Praxair 22

23 (Danbury, CT). All the other chemicals were of reagent grade. All the chemicals were used without further purification Synthesis of Poly(Hydroxyethyl Methacrylate) (PHEMA) Gels Poly(hydroxyethyl methacrylate) (PHEMA) gels were synthesized by free radical solution or bulk polymerization of the monomer with photoinitiation. Briefly, 2.7 ml of monomer HEMA and 10 µl of EGDMA were mixed with 2 ml of deionized (DI) water. The solution was purged by bubbling nitrogen for 10 min. 6 µg of photoinitiator (Darocur TPO) was added to the monomer mixture with stirring for 5 min and the resulting solution was immediately injected into a mold composed of two 5 mm thick glass plates separated by a plastic spacer. The spacer thickness was chosen to be either 0.1 or 0.2 mm. The mold was then placed on Ultraviolet transilluminiator UVB-10 (Ultra Lum, Inc.) and the gel was cured by irradiating by UVB light (305 nm) for 40 min. The gel was cut in square shaped pieces (about cm) and dried in air overnight for further use Drug Loading The drug was loaded into the gels either by directly dissolving the drug in the polymerizing mixture (direct entrapment) or by soaking the gel in an aqueous drug solution. Due to minimal solubility of DXA in water, it was loaded by direct entrapment or by soaking the gel in drugethanol solution. The drug concentrations in the loading solutions and the drug loading by direct entrapment were conducted under conditions that led to drug loadings comparable to those required for therapeutic applications. Before the soaking step, a square piece of PHEMA gel (about cm) was boiled in 200 ml of DI water for 30 min to remove the unreacted monomer. Next the gel was soaked in 3mL of drug solution. DXA was loaded by soaking in drug-ethanol solution for a period of 3 hours. During the soaking in drug-ethanol solution, the dynamic concentration in the ethanol phase was not monitored. The total amount of drug loaded 23

24 into the gel was estimated by determining the amount of ethanol uptake by the gel, and then multiplying it by the drug concentration in the ethanol solution. It was thus implicitly assumed that in view of the very high solubility of DXA in ethanol, absorption of drug on the PHEMA polymer can be neglected when the gel is soaked in ethanol. At the end of the loading stage the gel was taken out and dried in air overnight, and subsequently used for release experiments. During soaking of gels in aqueous DX and DXP solutions, the dynamic drug concentration in the DI water (or PBS) was monitored by measuring the absorbance spectra of the solution over the wavelength range of 220 to 270 nm with a UV-VIS spectrophotometer (Thermospectronic Genesys 10 UV). The loading step was conducted till equilibrium was reached. The total amount of drug loaded into the gel was determined by finding the total amount of drug-loss from the aqueous solution Drug Release Experiments The drug release experiments were conducted by soaking the square shaped gels (about cm) in 3 ml of DI water (or PBS) at room temperature. It is noted that the gels that were loaded with drug by soaking in aqueous solutions were directly transferred from the loading solution to the release solution, and so these were fully hydrated at the beginning of the release experiment. However, gels that had the drug loaded by direct entrapment or by soaking in drugethanol solution were dry at the beginning of the release experiment. The effect of this difference was shown to be minimal by comparing release profiles from dry and hydrated gels with the same drug loading (results not shown here). During the release experiments, the dynamic drug concentration in the DI water (or PBS) was monitored by measuring the absorbance spectra of the solution over the wavelength range of 220 to 270 nm. After equilibrium was reached, the gels were transferred to fresh 3 ml solution (DI water or PBS) and the process was repeated. These 24

25 two releases are refereed as the 1 st and the 2 nd release. In some cases even a 3 rd release was performed Effect of Gel Thickness on Loading and Release of Drug To determine whether the drug transport in gels is controlled by diffusion, the procedures described above were repeated with thin gels (0.1 mm thick). The drug loading and release procedures for the thin gels were identical to those described above except that the volume of solutions was reduced by a factor of 2 to ensure that the ratio of gel to fluid volume was kept the same Effect of Ionic Strength in Release Solution on Loading and Release of DXP Since DXP is a charged molecule, its transport may be affected by the ionic strength. To investigate the effect of electrostatics on DXP transport, the loading and release experiments for DXP were conducted at several different ionic strengths. In these experiments the ionic strength was changed by adding NaCl to the PBS buffer. DXP loading and release experiments were carried with three different NaCl-DXP solution in PBS corresponding ionic strengths of 594, 1022, and 1202 mm. Also experiments were conducted in PBS that has an ionic strength of 173 mm Conversion of UV-VIS Absorbance to the Corresponding Concentration of Drug The absorbencies of solution in both the loading and the release steps arise mainly from the drug, but there is some additional absorbance due to small polymer chains that continue to diffuse out from the PHEMA gels. The absorbance spectra of HEMA and dexamethasone (DX, DXA, and DXP) partially overlap, and thus to distinguish between the absorbance from HEMA and the drug, the measured absorbance spectra was deconvoluted by expressing the measured absorbance as Abs = α Drug + β Control (2-1) 25

26 where Abs is the measured absorbance spectra from 220 to 270 nm, Drug is the absorbance spectra of the drug in the same wavelength range at some arbitrary concentration, Control is the absorbance spectra of the solution in which PHEMA gel without any drug was soaked, and α and β are constants. These constants were obtained by finding best fit values minimizing the error between measured absorbance and calculated absorbance according to Eq.2-1 using the function fminsearch in MATLAB. The concentration of drug was finally obtained by multiplying α to the concentration corresponding to Drug. The above procedure assumes that the absorbencies of these components are simply additive and linear in concentration, and this was verified by conducting several experiments. Also the accuracy of the procedure described above was established by determining drug concentrations in solutions of known composition. The difference between the fitted and the measured absorbance spectra was typically less than 1% Determinations of Critical Micelle Concentration (CMC) of DXP Surface tension isotherm of DXP was measured at room temperature (about 23 C) by creating a pendant drop, digitizing the shape and then fitting it to the Young-Laplace equation by using the Drop Shape Analysis System DSA100 (KRÜSS). A concentrated solution of the DXP (40 mg/ml) in PBS was prepared and then diluted successively to mg/ml which is far below the expected CMC. 2.3 Results and Discussion Dexamethasone (DX) Loaded Gel Loading (by soaking in aqueous solution) and release studies of DX Partition coefficient The partition coefficient K is defined as the ratio of the drug concentration in the gel and the concentration in the aqueous phase at equilibrium. The values of partition coefficient can be 26

27 obtained from both the loading and the release experiments. For the loading experiment the partition coefficient can be calculated by K C g,f Vw(C w,i C w,f ) = = (2-2) C V C w,f g w,f where V w and V g are the volumes of the aqueous phase and the fully hydrated gel, respectively, and C g,f, C w,i and C w,f are the equilibrium concentrations of the drug in the gel, and the initial and equilibrium concentrations in the aqueous phase, respectively, in the loading experiment. For the 1 st release, the partition coefficient is given by K C g,f w w,i w,f r, f = = (2-3) C w,f V (C C V C g r,f C ) where C r,f is the equilibrium concentration of the drug in the aqueous phase gel in the 1 st release, and the other variables correspond to the values obtained from the loading phase for the same gel. Similar expressions can be written for the 2 nd and the 3 rd releases. Table 2-1 shows the equilibrium concentrations in the water phase and corresponding partition coefficients. The partition coefficient of DX seems to be constant in the whole concentration range that was explored in these experiments. Moreover, the K values are similar for the loading, 1 st release, and 2 nd release. This suggests that the process of loading and release of DX is reversible and that K is not function of concentration in the explored concentration range. The K values for loading and release were ± 3.14 and ± 3.41, respectively and mean K was Dexamethasone (DX) loading dynamics A schematic of the geometry of the gel used in the drug loading and release experiments is shown in Figure 2-2. The experimental data shown above demonstrates that the partition coefficient for dexamethasone is much larger than 1, which implies that a large fraction of the 27

28 drug is bound to the PHEMA polymer. The binding of the drug to the gel can be modeled as a Langmuir adsorption isotherm, which relates the adsorbed concentration of the drug on the gel (Γ) to the free concentration in the aqueous phase inside the gel (C) by the following equation Γ C Γ = k + C (2-4) where Γ is the surface concentration at the maximum packing on the surface and k is the ratio of the rate constants for desorption and adsorption of the drug on the HEMA surface. The mean concentration in the gel (C g ), which is essentially the sum of the bound concentration and the free concentration, is given by C S = Γ fc (2-5) V g + gel where S V gel is the surface area per volume available for the drug to adsorb and f is the volume fraction of water in hydrated gel. The value of f for PHEMA gels was determined to be 0.42 from the swelling experiments. At equilibrium the free drug concentration in the gel is expected to equal to the concentration in the PBS, and thus the partition coefficients determined above are simply the ratio of the total drug and the free drug, i.e., C g a K = + f (2-6) C k + C where a S V gel Γ. The values of K shown above are independent of C, which suggests that the value of k for DX adsorption on HEMA is much larger than mg/ml which is the highest equilibrium concentration investigated in this study. Thus K is treated as independent of concentration in the model developed below. The transport of the drug in the hydrogel is 28

29 expected to occur by a combination of bulk and surface diffusion, and thus it can be described by the modified diffusion equation, i.e., ( C t g ) = fd f 2 C 2 y 2 P Γ + Ds 2 A y (2-7) where D f and D s are the diffusivities of the drug in solution and on the surface, respectively, and P/A is the perimeter of the gel fibers per unit cross-sectional area, which can be approximated as S/V. Utilizing Eq.2-5 and Eq.2-6 in the above equation and noting the fact that K is independent of C gives C K = t 2 2 C ( fd f + Ds ( K f )) = D 2 2 where D ( fd + D ( K f )) f s y C y is the effective diffusivity of drug in the gel. The above (2-8) differential equation is subjected to the following boundary conditions C ( t, y) y y= 0 = 0 (2-9a) C ( t, y = H ) = (2-9b) C w where H is the half-thickness of the gel. The boundary condition Eq.2-9a arises due to symmetry at the center of the gel and the boundary condition Eq.2-9b assumes that the drug concentration in the water phase in the gel phase at the interface of gel and solution is the same as the concentration in the outer water phase. From a mass balance of drug in the aqueous phase, we get dcw Vw dt C = 2 Ag D (2-10) y y= H where A g is a cross-sectional area of the gel. The initial conditions for the drug loading are 29

30 = C ( t = 0, y) = 0 (2-11a,b) C w ( t 0) = C w, i The above set of equations was solved by implicit finite difference method with 21 spatial nodes and a dimensionless time step (D t/kh 2 ) of The DX loading experiments were conducted with different initial concentrations and the dynamic drug concentration in the aqueous phase was measured. These data was fitted to the model described above and diffusivity of the drug in the gel was determined. The error between the experimental data and model prediction was defined as 2 C w Cw, ex ) / Cw, ex (, where C w and C w,ex are the predicted concentration in the aqueous phase by model and the experimental concentration, respectively. The diffusivity for the each set of experimental dynamic concentration data was evaluated using the function fminsearch in MATLAB. The value of D for DX was ± m 2 /sec (n = 10). The theoretical profiles predicted by the model and the experimental data in drug loading experiments are given in Figure 2-3. Two different theoretical profiles for the same initial concentration correspond to the two different gel volumes. The model predictions match the experimental data very well with the error ranging from 1.86 to 3.88% Dexamethasone (DX) release dynamics The dynamics of drug release from a gel into fresh solution can be described by Eq.2-8 with boundary conditions Eq.2-9, Eq.2-10, and the following initial conditions C w ( t = 0) = 0, C ( t = 0, y) = Ci (2-11c, d) where C i is an initial free drug concentration in the gel. If the gel properties do not change in the drug loading experiments, the values of K and D for drug release are expected to be the same as during the loading step. The numerical procedure described above was repeated with the release data to determine the drug diffusivity in the gel. The experimental data and the fitted profiles are plotted in Figure 2-4. The amount of drug loading in the gel and the rms errors in the fit are also 30

31 indicated in the figure. The model predictions match the experimental release data well with error range between 1.60 and 3.29 %. The errors are slightly larger than those in drug loading, perhaps due to error accumulation to the previous drug loading experiments. From the fitting of the release data, the value of K and D for DX were determined to be ± 3.41 and ± m 2 /sec, respectively. These values are in excellent agreement with the values obtained from the loading data Release of DX loaded in the gel by direct entrapment Partition coefficient Drugs can be loaded to the hydrogel contact lenses by soaking the gels in drug solutions, or by directly adding the drug to the polymerizing mixture (direct entrapment). While direct entrapment of drug may be more convenient, there is a possibility that a fraction of the drug may get irreversibly trapped in the gel. To investigate the feasibility of loading DX by direct entrapment, gels with different drug loading were prepared and drug release experiments were conducted from these gels. The release experiments were conducted with four different initial drug loadings (M 0 /M g = 1.37, 2.74, 4.78, and 6.82 mg/g). Each drug-laden gel was soaked in 3mL of DI water (1 st release). After equilibrium was achieved, the aqueous solution was replaced with 3mL fresh DI water (2 nd release). The 3 rd release was conducted in the same manner. The equilibrium concentration in the water phase can be used to determine the gel concentration by a using a mass balance, and then these can be used to determine the partition coefficients. In Table 2-2, the equilibrium concentrations in the water phase and the corresponding partition coefficients are listed. The values of K in Table 2-2 are significantly higher than the K values obtained above. Moreover, the K values for the 2 nd release are much higher than those for the 1 st release. Both of these effects could possibly be caused due to 31

32 irreversible entrapment of a fraction of the drug. Based on this hypothesis, the mass balance in the drug-water system can be modified as M = V C + KV C = V C + K' V C + M ) (2-12) 0 w w, f g w, f w w, f ( g w, f p where M 0 is the total mass of drug loaded in the gel, M p is the mass of drug that is irreversibly trapped in the gel, K is the true partition coefficient of drug, and the other variables have the same definitions as above. The K values that were calculated by neglecting any irreversible entrapment are now called apparent partition coefficients ( K app ) to differentiate them from the intrinsic partition coefficients K. Based on the data obtained from the loading and release experiments from gels that had the drug loaded by soaking, it is reasonable to assume that the intrinsic partition coefficient is independent of concentration, and so for a given gel the K values and also the M p should be the same for the 1 st and the 2 nd releases. Using Eq.12 for both the 1 st and the 2 nd releases, M p and K can be computed for each gel, and these are listed in Table 2-2. The values of K (=34.30 ± 2.82) are in reasonable agreement with the values obtained from gels that were loaded by soaking in drug solutions. To further validate the results for K and M p, the apparent partition coefficient was determined by using the above equation for the 3 rd release and the calculated values were in good agreement with the measured values. The results also show that about 17.4% of total drug loaded (Table 2-2) is permanently entrapped in the gel matrix and not available for the drug delivery. The permanent entrapment could be either physical entrapment in tightly bound regions or chemical entrapment due to reactions Dexamethasone (DX) release dynamics The model for the release of drug directly entrapped in the gel into the water is the same as that for the release of drug loaded by soaking. The initial concentration in the gel, which is a parameter required for the fit is set equal to the concentration of the drug that is available to 32

33 diffuse. In Figure 2-5, the experimental data for 1 st and 2 nd release are plotted along with the model predictions. The mean values of D for 1 st and the 2 nd releases are ± m 2 /sec, and ± m 2 /sec, respectively. The value of D obtained by fitting the 1 st release data is in excellent agreement with the values reported above but the value increases by about 20% in the 2 nd release. This is an unexpected result because the drug release in the first release step is not sufficiently large to impact the structure of the gel Effect of gel thickness on DX loading A major fraction of the drug present in the gel is bound to the polymer and so the drug transport may occur by a combination of surface and bulk diffusion. If surface diffusion rates are small, the bound drug could desorb and then diffuse via bulk diffusion. To determine whether transport of drug is controlled by diffusion or by the process of adsorption-desorption of drug on the polymer, it was decided to fabricate gels with different thicknesses (0.1 mm and 0.2 mm thick in dry state). If the rate limiting step is diffusion, an increase in gel thickness by a factor of 2, while keeping fluid to gel ratio fixed, will increase the time scale of release by a factor of 4. A change in time scale by a factor of less than 4 signifies that the adsorption-desorption process occurs on a time scale comparable to the diffusion. As described in methods section, both thick and thin gels were cut into square shaped pieces (about cm) and unreacted monomer was removed by boiling. The gels were then dried and weighed and then soaked in water overnight for hydration. The hydrated thick gel (H = 0.13 mm) was put in 3 ml of DX-water solution and the hydrated thin gel (H = 0.06 mm) was put in 1.5 ml of DX-water solution for the drug loading studies. The drug loading and release profiles from the thick and the thin gels are shown in Figure 2-6. As mentioned above, if the drug transport is diffusion limited, the drug loading rate should 33