Modulating the Release Kinetics of Paclitaxel from Membrane-Covered Stents Using Different Loading Strategies

Membrane-covered Express2TM Monorail® stents composed of chitosan (CH) blended with polyethylene oxide (PEO) in 70:30% wt (CH-PEO) were coated with a monolayer of hyaluronic acid (HA). This significantly improved the resistance to platelet adhesion and demonstrated excellent mechanical properties, resisting the harsh conditions during stent crimping and subsequent inflation. CH-PEO/HA membrane was then combined with a paclitaxel (Pac) delivery system via three different approaches for comparison of release profiles of Pac. The activity of Pac in these systems was confirmed since its presence in the membrane significantly decreased cell viability of U937 macrophages. Presented results are promising for applications requiring different release patterns of hydrophobic drugs.


Introduction
Restenosis is an inflammatory response that may involve thrombus formation causing the renarrowing of a coronary artery. It occurs within 3 -6 months in 40 -50% of patients who have had angioplasty. This incidence was reduced to 20% with the use of bare metal stent implantation procedures, along with the administration of anti-proliferative, anti-migratory and anti-inflammatory drugs [1][2][3]. Consequently, drug-eluting stents (DES) have emerged as a more controlled and effective solution to inhibit the development of in-stent restenosis [4][5][6]. Stents coated with anti-proliferative and anti-inflammatory drugs such as paclitaxel, sirolimus, actinomycin D and tacrolimus acting as inhibitors of cell migration and cell cycle progression are currently under evaluation in several clinical trials [6,7]. Among these drugs, paclitaxel (Pac) has been proven to be one of the most promising agents for the treatment of neo-intimal hyperplasia [7,8]. Pac-eluting stents with the trade name of Taxus ® Express 2TM from Boston Scientific have received European and FDA approval and are currently used in most stenting procedures [8]. Nonetheless, the well-controlled and effective delivery of Pac remains a challenge. An intact drug-carrying membrane deems necessary for the functionality of the DES and the incorporation of drugs in a biodegradable polymer as a membrane coating has been suggested [9,10]. In a previous work, we have introduced a membrane-covered stent using a blend of chitosan (CH), a natural polysaccharide and polyethylene oxide (PEO) as membrane material [11]. CH is a linear cationic polymer of D-glucosamine obtained by alkaline N-deacetylation of chitin that has been extensively studied in recent years [12]. Due to inter-and intra-molecular hydrogen bonding, CH features excellent film forming properties and high mechanical strength suitable for membrane formation [12]. CH matrices of various geometries, pore sizes and orientations can be formed either alone or after blending with various macromolecules such as PEO, a synthetic and neutral watersoluble polymer [12,13]. The permeability of solutes through the membranes prepared by blending CH with PEO is much higher than that through CH alone [14,15]. Therefore, CH-PEO appears to be suitable for improving the permeability of toxic metabolites and reducing thrombogenicity. Amiji et al. demonstrated that the permeability coefficient of urea increased from 5.47 × 10 -5 cm 2 ·min -1 in CH to 9.89 × 10 -5 cm 2 ·min -1 in CH-PEO membranes [15]. PEO has also been shown to effectively increase the blood compatibility of polymeric materials including CH [15]. Biomaterials grafted with PEO are able to resist plasma protein adsorption and platelet adhesion predominately by a steric repulsion mechanism [16]. To reduce plasma protein adsorption, platelet adhesion and activation, and thrombus development, the surface of CH membranes can be modified by the electrostatic interaction of cationic CH and anionic polysaccharides such as hyaluronic acid (HA) [11,17]. Drug-eluting membranes consisting of chitosan/hyaluronan multi-layers show promising properties for use as a stent coating material [11,[17][18][19][20]. HA is a high molecular weight glycosaminoglycan found in the extracellular matrix of arterial smooth muscle cells and endothelial cells and have shown good blood compatibility [21]. A number of strategies for the modification of HA mainly through the carboxyl and hydroxyl groups have been developed including esterification and carbodiimide chemistry [18,22]. HA has been used to cover stents via cross-linking with N-(3-dimethylaminopropyl)-N′-ethylcarbodiimide (EDC) leading to a reduced inflammatory response compared to un-coated stainless steel stents in pig coronary arteries [23]. Additionally, it significantly reduced platelet deposition by 55% after 2 hours of blood exposure in exteriorized arteriovenous shunts in baboons [21].   The release was observed over a period of 65 hours. In vitro release of physically incorporated Pac from CH-PEO/HA membranes showed that 0.233 mg ± 0.014 mg, 0.522 mg ± 0.036 mg and 0.586 mg ± 0.025 mg of Pac was released within 65 hours for 0.24 mg, 0.51 mg and 0.60 mg, respectively. The release of drugs from matrix systems can be described with the Higuchi equation [26]. The basic equation of the Higuchi model is: where M t is the cumulative absolute amount of drug released at time t, A is the surface area of the controlled release device exposed to the release medium, c 0 and c S are the initial drug concentration, and the solubility of the drug in the polymer, respectively, and D is the drug diffusivity in the polymer carrier [26,30]. This model considers diffusion as the dominating mechanism for drug release with the proportionality between the cumulative amount of released drug and the square root of time is an indicator for diffusion controlled drug release systems. The slope K, the axis intercept a and the squared coefficient of correlation R 2 from linear regression were calculated according to equation (2): where M ∞ is the total cumulative amount of drug released at infinite time t and K is a constant characterizing the design variables of the system [30]. Table 1 illustrates the resulting values from linear regression based on equation (2). A plot of released Pac against the square root of time showed good linearity for all experiments.        In vitro release of Pac from microspheres embedded into CH-PEO/HA membranes showed that 0.264 mg ± 0.008 mg and 0.328 mg ± 0.008 mg of Pac was released within 216 hours for the 8.0 mg and 9.5 mg experiment respectively, which is 61.7 % and 64.6 % of the initially incorporated drug in each case. Table 2 demonstrates the linear regression results based on equation (2) for the release of Pac from microspheres embedded into the CH-PEO/HA membrane. Negative axis intercepts are attributed again to a lag effect particularly occurring when hydrophobic drugs are released into an aqueous medium [31]. Table 2. Slope K, axis intercept a, squared coefficient of correlation R 2 and the plot from equation (2) for the release of Pac (over time in hours) from microspheres embedded into CH-PEO/HA membranes in phosphate-buffered ethanol:water (40:60 v/v) solution.
The calculated squared coefficients of correlation R 2 from the release of Pac out of microspheres embedded into the membrane were lower if compared to the release of physically incorporated Pac. The used PLGA polymer for the preparation of microspheres is cleaved into shorter chain alcohols and acids upon contact with water [33]. The Higuchi equation is based on a diffusion controlled system [26,30] and thus polymer degradation contribution to the release mechanism of Pac from PLGAmicrospheres explains the lower K values in this system if compared to the coefficients found for the release of physically incorporated paclitaxel. The difference in the reported values is not amplified in the release profile due to the variation in the release time. The use of PLGA offers various important advantages over other controlled release systems, such as (a) the possibility to control the release rate over a period of days or months, (b) good biocompatibility and (c) complete biodegradability [33]. PLGA based microparticles are known to be bulk eroding, because water penetration into the system is much faster than the subsequent polymer chain cleavage [33]. The eroding microspheres have to remain in the membrane during degradation to avoid flushing of particles into the bloodstream. Compared to the release of physically incorporated Pac, the drug release from microspheres resulted in significantly slower release. Furthermore, cell culture tests confirmed the activity of low amounts of

Thrombogenecity
Blending CH with PEO improved the resistance to platelet adhesion (2.27 ± 0.16 10 6 platelets/cm 2 vs. 1.92 ± 0.15 10 6 platelets/cm 2 for CH and CH-PEO respectively), as shown in Figure 11. Further improvement was achieved by coating the CH-PEO membrane with a monolayer of HA which led to a significantly reduced platelet adhesion compared to CH alone (1.25 ± 0.17 10 6 versus 2.27 ± 0.16 10 6 for CH, p<0.002). Hydrogels are highly hydrated (>95% w/w) and exhibit smooth surfaces counteracting cell attachment [34]. Additionally, recent studies have indicated that the high anionicity of native long chain HA poorly interact with cells in vitro and show poor platelet binding characteristics which is likely due to their extreme hydrophilicity and anionic surface charge. It seems that the surface functional groups of HA and their charge characteristics play an important role in platelet adhesion [34].

Preparation of hyaluronan (HA) ester prodrug of paclitaxel (Pac)
The HA ester prodrug of Pac (HA-Pac) was prepared in three steps as reported elsewhere [18,25]. Briefly, 2'-Succinyltaxol was prepared and Pac was linked to HA via a labile succinate ester linkage [25]. Subsequently the 2'-succinyltaxol was converted into an activated ester using carbodiimide coupling chemistry with N-hydroxysuccinimide (NHS) and dicyclohexylcarbodiimide (DCC). The activated ester was linked to an amine modified HA. The last purification steps in the synthesis of HA-Pac were dialysis of the HA-Pac solution against acetone/water (70:30) and water (membrane tubing, molecular weight cut-off 50000). Pac loading was determined by UV absorbance (λ max = 227 nm, ε = 2.8 × 10 4 ) in 80:20 CH 3 CN:H 2 O.

CH-PEO membrane formation
CH-PEO blend was used for film preparation and for the membrane-covered stent. CH was dissolved in 0.1 M aqueous acetic acid (2% w/w) and then filtered. PEO dissolved in glacial acetic acid was then added to the CH solution to form a CH/PEO blend with a ratio of 70:30 by weight. The blend was degassed and either used for the formation of films or for the preparation of the membrane covered stent. The dried material was washed with 0.1 N NaOH to de-protonate the ammonium functions of the CH. Subsequently, the membrane was washed thoroughly with PBS buffer and finally with 0.14 M aqueous NaCl. Microsphere-containing membranes were prepared by suspending the microspheres into the CH-PEO blend. All further steps were carried out as mentioned with CH-PEO membrane formation.

Membrane-covered stent preparation
The membrane-covered stent preparation was carried out according to a modified procedure reported by Thierry et al. [11]. CH-PEO (70:30) blend solution was used for an Express 2TM Monorail ® stent system. The stent was mounted on an inflated rotating balloon catheter and the degassed CH-PEO solution was applied onto it. Following overnight drying, the balloon was deflated and the stent was immersed into a 0.1 N NaOH solution.

Self-assembled monolayer coatings: membrane-HA and membrane HA-Pac preparation
A solution of sodium hyaluronate (1 mg/mL in 0.14 M aqueous NaCl) and a solution of the HA prodrug of paclitaxel (HA-Pac 1 mg/mL in 0.14 M aqueous NaCl) were prepared separately. Ultrapure water (UPW) was used in all experiments (18.2 MΩ cm 2 ; MilliQ system, Millipore). The monolayer build-up was accomplished by applying the polymer solution (HA or HA-Pac) onto the CH-PEO substrates. The adsorption time was 5 minutes followed by a washing step with 0.14 M NaCl.

Membrane assessment during inflation
The membrane-covered stent was hand-crimped onto the balloon catheter [Express 2TM Monorail ® stent system on 5F Guide Catheter/minimum I.D. (0.059″/1.47 mm)]. To simulate the friction generated during the application of a stent via a balloon catheter inside a blood vessel and to assess the membrane integrity during inflation of the CH-PEO/HA covered stent inside the 2mm tube, the procedure was monitored with a microscope through a quartz cuvette which was attached to the tubing. The cuvette and the tube were filled with PBS buffer solution and the inflation was carried out at 37°C.

Preparation of microspheres
PLGA microspheres were formulated according to the o/w-solvent evaporation method. 1 mL of PLGA solution (50 mg/mL) with 10 % paclitaxel (w/w) in methylene chloride was poured into an aqueous PVA solution (4% w/v, 10 mL). After stirring, the formed emulsion was treated with ultrasound for 5 minutes and subsequently transferred into a 500 mL round bottom flask. The organic phase was removed under reduced pressure. The remaining spheres were centrifuged for 5 min at 345g and the PVA supernatant was removed carefully. The spheres were re-suspended in UPW and ultrasonicated for 5 minutes. Centrifugation, exchange of the supernatant with UPW and sonication were repeated three times to remove the PVA. The microspheres were dried then by lyophilization. Particle size measurements were carried out using dynamic light scattering (DLS) via a low angle laser lightscattering device (Malvern Instruments HPPS).

SEM analysis
Un-coated and CH-PEO/HA membrane-coated Express 2TM stents were sputtered with an ultra-thin layer of Au-Pd. Microscopic imaging using a Field Emission Gun Scanning Electron Microscope (FEGSEM; Hitachi model S-4700) at an accelerating voltage of 5 kV was done.

Determination of drug loading
Drug loading was determined by dissolving accurately weighed amounts of microspheres in dichloromethane and subsequent drug detection at λ max = 229 nm (UV µ-Quant, Bio-Tek Instruments, Inc., VT, USA).
Macrophages (10 5 cells/mL) cultured in suspension were exposed to different surfaces. Macrophages alone served as negative control. 100,000 cells per mL were exposed to the CH-PEO/HA membrane, CH-PEO/HA/Pac membrane with chemically linked Pac and to the CH-PEO/HA with physically incorporated Pac. The UV-spectrum of paclitaxel in an ethanol/water solution showed a maximum extinction at λ max = 229nm whereas an aqueous solution of HA did not show any absorption at that range. The synthesized bioconjugate of Pac and HA showed a maximum at 228 nm indicating the successful linkage of Pac to HA. The loading of Pac on the polymer backbone was quantified by UV absorbance at 228 nm. HA-Pac with an amount of 10% Pac was used in all following experiments. Incubations were conducted at 37°C in a humidified 5% CO 2 environment. After 24 hours the cells were stained with a trypan blue solution and both, the number of stained cells and the total number of cells were counted using a hemocytometer under the microscope.

Drug release study from CH-PEO/HA + PLGA membranes
Isolated Pac-loaded microspheres (8.0 and 9.5 mg, respectively) were added to a CH-PEO solution, cast into 24-well cell culture plates and dried over night. The dried membranes were de-protonated with 0.1M NaOH, neutralized with PBS buffer, and subsequently coated with HA (1 mg/mL in 0.14 M aqueous NaCl). Drug release studies were carried out similarly as described earlier.

Isolation and labelling of platelets
Platelets were isolated and radio-labelled with 51 Cr as described previously [27]. A human blood sample collected in acid citrate dextrose was centrifuged at low speed to obtain platelet-rich plasma. Incubation with the radioactive 51 Cr (Amersham International) followed. The suspension was centrifuged to remove unbound 51 Cr and re-suspended in platelet-poor plasma. 245x10 6 platelets were found per mL plasma. Platelet adhesion experiments were carried out with different polymer foils and polymer-composites for 1 hour at room temperature. The size of the foils was 1 cm × 0.5 cm resulting in an exposed area of 1 square centimetre (neglecting the height of the foils). Subsequently, the samples were placed in a gamma counter equipped with a nuclide analysing program for 51 Cr radioactivity determination. The amounts of platelets per square centimetre adhered to the samples was calculated from the samples radioactivity and the cellular count of reference calibration. The calibration line showed a good linearity (R 2 = 0.9996) indicating measurement accuracy.

Statistics
Results are expressed as mean values ±standard deviation (SD). Paired student t-tests were used and ρ<0.05 was considered statistically significant.

Conclusions
In this work, we have reported a membrane-covered stent from CH-PEO blends with attractive mechanical properties and enhanced drug loading and release capacity. We showed that the combination of the hemocompatible CH-PEO/HA membrane with a drug release system is feasible and that drug release kinetics can be modulated using the desired loading strategy. The absence of surface irregularities due to the HA coating shows the potential for abolishing platelet deposition and thus thrombus formation. The membrane offers the possibility of incorporating the anti-proliferative agent paclitaxel using various strategies: (i) surface bound paclitaxel via hydrolysable ester linkage of drug and HA, (ii) physical incorporation of paclitaxel into the membrane, and (iii) PLGA microspheres loaded with paclitaxel and embedded into the membrane to modulate drug release. Significantly slower release of paclitaxel from PLGA microspheres (~ 60% in 216 hours) could be achieved if compared with the release from physically incorporated or covalently-bound paclitaxel (~ 100% in 65 and 43 hours, respectively). In either membrane system, the Pac-loaded membranes inhibit the viability of proliferating cells such as U937 human macrophages as a result of drug release and demonstrate improved thrombogenicity due to a monolayer coating of the membrane with HA. Thus, it is promising for the clinical administration of highly hydrophobic drugs to more than one application requiring different release patterns such as in the treatment of aneurysms as well as in rare and lifethreatening complications of sealing coronary perforations.