Drug Release Properties of Polymer Coated Ion-Exchange

Drug Release Properties of Polymer Coated Ion-ExchangeResin Complexes: Experimental and Theoretical Evaluation

SEONG HOON JEONG,1 NAHOR HADDISH BERHANE,2 KAMYAR HAGHIGHI,2 KINAM PARK1

1Department of Pharmaceutics and Biomedical Engineering, Purdue University, West Lafayette, IN 47907

2Department of Agricultural and Biological Engineering, Purdue University, West Lafayette, IN 47907

Received 19 September 2005; revised 18 April 2006; accepted 21 April 2006

Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/jps.20677

ABSTRACT: Although ion-exchange resins have been used widely as drug deliverysystems, their exact release kinetics has not been reported yet. Usually only the rate-limiting step has been taken into account and the rest of the steps have been ignored asinstantaneous processes. To investigate the exact release kinetics of polymer-coateddrug/ion-exchange resin complexes for sustained drug delivery, the results of newmathematicalmodelingwere comparedwith experimental results.Drug/resin complexeswith a model drug, dextromethorphan, were prepared and used as cores for fluid-bedcoating. Anaqueous colloidal dispersion of poly(vinyl acetate)was applied for the coating.A comprehensive mathematical model was developed using a mechanistic approachby considering diffusion, swelling, and ion-exchange processes solved by numericaltechniques. The rate-limiting factor of the uncoated resin particles was diffusionthrough the core matrix. Similarly, in the coated particles the rate-limiting factorwas diffusion through the coating membrane. The mathematical model has capturedthe phenomena observed during experimental evaluations and the release dynamicsfrom uncoated and coated (at different coat levels) particles were predictedaccurately (maximum RMSE 2.4%). The mathematical model is a useful tool totheoretically evaluate the drug release properties from coated ion-exchange complexesthus can be used for design purposes. � 2006 Wiley-Liss, Inc. and the American Pharmacists

Association J Pharm Sci

Keywords: ion-exchange resin; drug delivery; mathematical modeling; diffusion;fluid-bed coating

INTRODUCTION

Some of the early applications of ion-exchangeresins were in the field of chemical engineering forpurification of fluids from contaminants andseparation of different gases. Since then, ion-exchange resins have been used in the pharma-ceutical industry as drug delivery systems. Theirapplications range from tablet disintegrants,

taste-masking, and stabilization to key compo-nents of extended release formulations.1–3 Somedrug products containing drug-loaded resins havebeen introduced in the market.

For the purpose of drug delivery, the drugrelease rate of ion-exchange resins can bemodifieddepending on functional groups, ion exchangecapacity, degrees of cross-linking, and particlesize. However, the release rate from the resincomplexes may not be satisfactory for sustainedrelease. Thus, further control of the release ratehas been achieved by applying a permeable coat-ing.4–8 This implies that the drug release can becontrolled by one or any combination of the core,coating and film diffusion resistances.

JOURNAL OF PHARMACEUTICAL SCIENCES, 2006 1

Correspondence to: KinamPark, Professor (Telephone: 765-494-7759; Fax: 765-496-1903;E-mail: [email protected])

Journal of Pharmaceutical Sciences, (2006)� 2006 Wiley-Liss, Inc. and the American Pharmacists Association

Ion-exchange resins possess different releaseproperties. They have fixed ionic functional groupswhich can provide binding of ionic drugs. Releaseof the bounddrugs requires exchangewith counterions such as hydrogen or sodium, which areavailable in the gastrointestinal tract. The overalldrug release kinetics of the polymer coated ion-exchange resins are mainly dependent upon thedrug and counter ion diffusion resistance in thecoating film and the boundary layer (film) sur-rounding the particles. The dissociation of drug atthe reaction surface is very fast as compared toother pertinent processes and barely does it affectthe overall drug release kinetics.

To elucidate the drug release mechanisms andrate controlling steps in coated pellets or particles,considerable mathematical modeling efforts havebeen carried out for pharmaceutical9–16 as well asagricultural17,18 applications. These studies dealwith coated pellets with either granular solubledrug or matrix loaded with a soluble drug inthe core. The processes involved in these systemsare liquid diffusion, drug dissolution, and drugdiffusion within the pellet and surrounding med-ium. In these studies, analytical derivations of thediffusion-dissolution equations were examined byapplying simplifications and approximations. Tomake these equations amenable for analyticalsolving, usually only the process-limiting step istaken into account ignoring the dynamics of somepertinent processes. In fact, Frenning12 hasrecently reported that only a few consider dynamicdescriptions of the pertinent processes as contri-buting steps to drug release from coated pellets.Moreover, these models can only be used forhomogeneous dosage forms where the properties(notably diffusion coefficients) of the differentdomains (such as core and coating) are lumpedwhich limit their applicability for design purposes.However, with the application of powerful numer-ical techniques, dynamic description of pertinentprocesses can be considered with less approxima-tions and simplifications that can result not only ina more accurate prediction of the drug releaseprofiles but also makes the models useful fordesign purposes such as selection of particle typeand size, and coating type and thickness.

The objective of this study was to investigatedrug release properties of polymer coated ion-exchange resin complexes using experimentalevaluation and to formulate a mechanistic drugrelease model by considering dynamic descriptionof pertinent processes (diffusion ion-exchange,swelling) using numerical approach. It is believed

that such modeling approach will lead to accurateprediction of the drug release by using physicallymeaningfulmodel parameters instead of empiricalones. Such a tool is important for further under-standing release mechanisms and deign purposesleading to renewed interest in the ion-exchangeresin system for taste-masking and sustaineddrugrelease for new fast-melting tablet formulations ofvarious ionic drugs.

EXPERIMENTAL

Materials

An ion-exchange resin (Dowex1 50WX4-200,polystyrene sulfonate, Hþ form, 4% divinylben-zene content) was purchased from Sigma-Aldrich(Sigma-Aldrich, Inc., St. Louis, MO). The resinparticles were purified by rinsing 200 g of wetresin three times with 1000 mL of distilled water,twice with 95% ethanol, and then twice with1000 mL of distilled water to remove the ethanol.Each treatment took at least 8 h by a batchprocess. After filtration, the resin was dried in a458C oven, and the moisture content was evalu-ated using a Karl Fischer titrator (Model 270,Denver Instrument, Arvada, CO). Dextromethor-phan hydrobromide monohydrate (DM), a modeldrug, was obtained from Spectrum1 (SpectrumChemical Mfg. Corp., New Brunswick, NJ), andKollicoat SR1 30D (polyvinyl acetate aqueousdispersion) was donated by BASF (BASF Corp.,Mont Olive, NJ). Kollicoat1 SR 30D is known asone of the aqueous colloidal polymer disper-sions for the manufacturing of pH-independentsustained release drug delivery systems. The poly-mer dispersion is based on polyvinyl acetate (27%w/w) stabilized with polyvinylpyrrolidone (2.7%w/w) and sodium lauryl sulfate (0.3% w/w).

Preparation of DM-Loaded Resin Complex

The DM-loaded Dowex1 50WX4-200 complexwas prepared by a modified batch process. Thepurified resin particles were dispersed in a 1.9%w/v drug solution with magnetic stirring atroom temperature for 3 h. After decanting theclear supernatant carefully, the same volume offresh drug solution was added and stirred againfor 5 h at room temperature. The complex wasseparated from the supernatant by vacuumfiltration, washed with deionized water to removeany uncomplexed drugs, and then dried in the

2 JEONG ET AL.

JOURNAL OF PHARMACEUTICAL SCIENCES, 2006 DOI 10.1002/jps

oven. The drug content of the resin complex wasexamined.

Preparation of Polymer Coated Resin Complexes

The DM-loaded resin complex was coated withKollicoat SR1 30D in a fluidized-bed coater, MFL-01 (Vector Corporation, Marion, IA) to obtain apredetermined weight gain. Bottom spray coatingmethod (Wurster process) was applied for thisprocess. The dried resin complex (40 g) was mixedwith micronized talc (0.8 g) to improve the initialflowability before the coating process. The coatingsolution was diluted to 10.0% w/w solid content.In order to enhance the film formation and theflexibility of the films, plasticizer (triethyl citrate)was added. Formulations of the coating solutionand operating conditions of the fluid-bed coaterare shown in Table 1. The maximum amount ofcoating applied was 20% w/w as compared withthe amount of core (resin complex).

Drug Release Test

Drug release tests from the uncoated and coatedresin complex particles were conducted accordingto USP 27 Apparatus 2 guidelines (the paddlemethod) (Vankel1 VK 7000, Vankel, Edison, NJ)with 900 mL dissolution medium maintained at37� 0.58C and mixed at 100 rpm. The dissolutionmedia used in this study were 0.1, 0.01, 0.001 NHCl. Samples were withdrawn at predeterminedtime intervals and analyzed for drug contentusing HPLC system (Agilent 1100 Series, AgilentTechnologies, Waldbronn, Germany) at a wave-length of 280 nm. Samples were filtered with0.2 mm nylon filters, and then 20 mL of the samplewas injected. The column used for the analysis

was a Symmetry1C18 5 mm (3.9� 150mm) (WatersCorporation, Milford, MA) with SentryTM guardcolumn (Symmetry1 C18 5 mm, 3.9� 20 mm). Themobile phase contained amixture of aqueous buffer(10 mM KH2PO4 adjusted to pH 2.6 withphosphoric acid) and acetonitrile in a volumeratio of 26:74. The retention time of DM was3.1 min and every standard calibration curve wasmade before analysis to monitor the linearity from0.5 to 250 mg/mL.

Determination of DiffusionCoefficients Through the Polymer Film

DM permeation studies through the cast filmswere performed using side-by-side diffusion cells(PermeGear, Inc., Bethlehem, PA). Polymer filmswere prepared from coating solution by casting ona Teflon surface. The casting molds were preparedin house. The films were dried in an oven for 24 hat 608C and then stored for 1 h at roomtemperature and 60% relative humidity. Thedried films were peeled off the Teflon surfaceand cut into pieces with the size enough to coverthe effective diffusion area of the cells. Thethickness of the films was determined using amanual micrometer (Mitutoyo Corp., Kawasaki,Japan) with an accuracy of �1 mm by measuringat least at seven random positions on the films.The cast films were placed between the twodiffusion chambers. For the diffusion coefficientof DM through the film, one compartment wasfilled with 3.0 mL of DM-saturated solution withexcess drug particles, and the other compartmentwas filled with 3.0 mL of 0.1 N HCl. For thediffusion coefficient of counter ion (Hþ) throughthe film, one compartment was filled with specificamount of drug loaded resin complex particlessuspended in 3.0 mL water and the othercompartment was filled with 3.0 mL of 0.1 NHCl solution. The exposed surface area was0.636 cm2. The volume of each compartmentwas 3.0 mL, and the temperature of the systemwas maintained at 378C throughout the experi-ments. An equal volume of fresh buffer wasimmediately added to the side where samplingoccurred. DM concentrations were quantified byHPLC analysis and corrected for dilution bysampling. The effect of compound withdrawal wastaken into account when calculating the accumu-lated amount of the drug permeated. Polymer filmswith three different thicknesses were evaluated forthis test. Diffusion coefficients (cm2/s) of DM andcounter ion through the polymer can be obtained

Table 1. Formulations and Operating Conditions forthe Preparation of Coated Particles

Conditions Value

Coating SolutionKollicoat SR1 30D (g) 70.00Triethyl Citrate (g) 1.05Water (g) 149.45

Operating ConditionsInlet Air Flow (l/m) 42.0Inlet Air Temperature (8C) 60.0Exhaust Temperature (8C) 24.3Nozzle Pressure (psi) 15.7Pump Speed (rpm) 10.0

POLYMER COATED ION-EXCHANGE RESIN COMPLEXES 3

DOI 10.1002/jps JOURNAL OF PHARMACEUTICAL SCIENCES, 2006

when plotting the flux and the concentrationgradient and the values beyond the lag time wereaveraged. For the diffusion coefficient of thecounter ion, it was assumed that bound drugespecially in the surface of the particles could bereleased as soon as the counter ion passedthrough the membrane.

Swelling Kinetics of Ion-Exchange Resins

Measuring the volume change of swollen particlesis a simple way to characterize the swellingkinetics and swelling equilibrium. Swellingkinetics of coated and uncoated resin complexeswere measured using luer-lock and non-jacketedglass liquid chromatography column (size: 1.0 cm�20 cm, bed volume 16 mL, Sigma-Aldrich). Acertain amount of the coated and uncoated resincomplex particles was transferred and packed tothe glass column equipped with a coarse fritted-glass disk at the bottom. Aqueous mediumwas pumped upward using a peristaltic pump(Minipuls 2, Gilson, France) to wet all theparticles at room temperature. Volume changeof the particles in the column was monitored atpredetermined time intervals and the swellingratio (Rs) can be calculated as follows:

Rs ¼Vs � Vd

Vdð1Þ

where Vs and Vd are the volume of swollen anddried resin particles, respectively.

Scanning Electron Microscopy (SEM)

Morphologies of DM-loaded resin complex andcoated resin particles were examined by the SEM.Dried samples were attached to specimen stubsusing double-sided copper tape and sputter coatedwith gold-palladium in the presence of argon gasusing a Hummer I sputter coater (Anatech Ltd.,Denver, NC). The samples were imaged with aJEOL JSM-840 scanning electron microscope(JEOL USA Inc., Peabody, MA) using 5 kVaccelerating voltage, 26�28 mm working dis-tance, and probe current of 3� 10�11 amp.

Confocal Laser Scanning Microscopy (CLSM)

The internal structure of coated resin particleswas imaged using an MRC-1024 Laser ScanningConfocal Imaging System (Bio-Rad Laboratories,Hercules, CA) equipped with a krypton/argonlaser and a Nikon Diaphot 300 inverted micro-

scope. All confocal fluorescence pictures weretaken with a 20� objective. Nile Red, a lipophilicfluorescence dye, was added to the coating solu-tion to visualize the coating film and the coreappeared blue due to the self-fluorescence ofthe DM. Red and blue fluorescence images andtransmission images were obtained from separatechannels.

THEORETICAL FRAMEWORK

Several semi-empirical models for drug releasefrom different drug delivery formulationsexist. Among them the power models (Higuchi,19

Korsemeyer,20 and Peppas21) and the Weibulmodels (Boyd,22 Reichenberg,23 and Bhaskar5)can be mentioned. A review of the differentmodels is given in Costa and Sousa Lobo24 Inthe present study, the well-known model for ion-exchange, the Boyd22 model which was latermodified by Reichenberg23 and Bhaskar5 wasemployed to conduct preliminary study and tovalidate some of the assumptions of the detailedmathematical model. Then a more comprehensivenumerical model was developed and validated.

Boyd Models

According to Boyd et al.22 the drug release fromion exchange resinate can be controlled by twokinds of diffusion processes, namely the diffusionof the drug across thin liquid film at the peripherytermed as film diffusion and the diffusion of thedrug in the matrix termed as particle diffusion.The rate-controlling step is either diffusion ofdrug across a thin liquid film at the periphery ofthe resin particle or diffusion of freed drug in amatrix. Both diffusion mechanisms are sequentialsteps so the slower one would be rate controlling.Assuming diffusion of a drug in the matrix is therate-controlling step and all the resin particlesare uniform spheres of radius r, the fraction ofdrug released, F, is given by the followingexpression.22

F ¼ Mt

M1¼ 1� 6

�2S1

n¼1

e�n2Bt

n2where B ¼ �2Di=r

2

ð2Þ

Mt andM1 are the amounts of drug released aftertime t and after infinite time, respectively.B is therate constant,Di represents the effective diffusioncoefficient of the exchanging ions inside the resinparticle, and n is the summation variable. Since

4 JEONG ET AL.


using Eq. (2) is not convenient due to its infiniteterms, some simple approximations were soughtsuch as that by Reichenberg23 (Eq. (3)).

Bt ¼ 2� 1��6 F � 1� �

3F� �0:5� �

forF � 0:85

�lnð1� FÞ � 0:04977 for F > 0:85

(ð3Þ

If the plot of the Bt corresponding to the F againsttime gives a straight line with a slope equal toB, itcan be assumed that drug diffusion within theresin matrix is the rate-controlling step and thediffusivity can be obtained using the Bvalue.4,6,8,25 Bhaskar proposed a more elegantapproximation based on the monotonic transformmethod:

� lnð1� FÞ ¼ 1:596

r

� �1:3

ðDitÞ0:65 ð4Þ

Hence, particle diffusion control can be tested bythe linearity between �ln(1�F) and t0.65 with theslope being a quantity related to the diffusivityand particle radius.

For film-diffusion controlled process Boydet al.22 proposed a model based on the assumptionthat the film thickness and distribution coefficientare constant in each resinate particle.

�lnð1� FÞ ¼ 3Pt

rð5Þ

where P is the apparent permeability of the film.The plot �ln(1�F) versus t provides a linear linewith a slope related to the apparent permeability,P.

Mechanistic Model

The processes involved during the release phaseare: (i) diffusion of the counter ion through theboundary layer, coating, and resin complex,

(ii) dissociation (ion exchange reaction), and (iii)diffusion of freed drug through the core and/orcoating. Since the drug release kinetics of thedelivery system is directly correlated with thefunctionality of individual particles, it is acommon practice to consider a single particle forthe mathematical treatment. The mathematicalmodel was formulated based on the followingassumptions. (i) The effects of convection andpressure gradients can be ignored. (ii) The pelletswere considered spherical in shape, based on theSEM and CLSM images shown in Figures 1A, 3and 4. (iii) The boundary layer on the surface ofthe coating was ignored due to the hydrophilicnature of the drug and existence of vigorousmixing, and a perfect sink condition with equili-brium counter ion concentration on the surfacewas considered. Note that this assumption rulesout the possibility of film diffusion as a controllingmechanism a priori which should be confirmedfrom the preliminary investigation using simplermodels. (iv) The pellets were assumed to behomogeneous and isotropic. As a result of thisand the foregoing perfect sphere assumption, asymmetry condition was invoked resulting in a2-D model of a single pellet. One quarter of thegeometric model of a typical pellet as consideredfor the simulation is shown in Figure 1B.

Given these assumptions the processes cangenerally be described by the reaction-transportequation.

@ci@t

¼ rJi þ Rx ð6ÞWhere c is the concentration of the diffusingspecies in the pellet and the subscript i denotesthe species (i¼ 1 for the water, i¼ 2 for thecounter ion and i¼ 3 for the drug); t, the time;Ji, the net flux; and Rx, the reaction source term.

Figure 1. Scanning electron micrograph of DM-Dowex1 50WX4-200 resin particles(A) and a schematic diagram showing the geometric model of a quarter polymer coatedresin complex (B).



Themost important feature that distinguishes ionexchange from isotopic exchange is the electriccoupling of the ionic fluxes. This means thediffusion process is governed by the Brownianmotion and the effect of the electric charge on themovement of the ions. Here, it should be notedthat the water does not take part in the electriccoupling of the ions and is considered here forthe effect of its (water) ingress on the swellingof the pellets. The resulting flux of the species(i¼ 2, 3) can be described by the Nernst–Plankequation.26

Ji ¼ ðJiÞdiff þ ðJiÞel

¼ �Di

�rci þ zici

F

RTrW

�; i ¼ 2; 3

ð7Þ

whereD is the diffusion coefficient; zi, the valence;F, Faraday constant; R, gas constant; T, absolutetemperature; W, electric potential.

Assuming that there are no co-ions present andthe swelling of the resin is neglected, the restric-tions of electro-neutrality (

Pzici ¼ const) and

absence of electric currentP

ziJi ¼ 0 apply.Applying these constrains on Eq. 7 yields:

Ji ¼ DDIrci ð8Þ

with

DDI ¼Q

Dið ÞP

zicið ÞPz2i Dici

¼P

zicið ÞPz2i

ciDi

ð9Þ

The reaction source term Rx can be described as:

Rx ¼ kYn

ci ð10Þ

where the subscript n represents the reactantspecies, in this case, the counter ion and the drug-resin complex; and k is the reaction rate constant.The physical interpretation of the reaction con-stant has little in common with the rate constantsin the actual chemical reaction; rather it is apartition coefficient describing how the ions arepartitioned between the phases. Moreover, thereaction term is constrained by the availabilityof the resin-drug complex and once depleted theterm disappears. Due to the speed of the dissocia-tion, chemically controlled process was excludedunequivocally as a rate controlling mechan-ism.22,26 Thus, the reaction term can be ignoredand included here only for the sake of complete-ness. Note that these equations only apply tothe core matrix. For the coating, only Fickiandiffusion occurs and the reaction source term doesnot apply.

To account for the swelling of the core of thepellet a moving boundary problem was consideredusing the Arbitrary Lagrangian-Eurelian (ALE)approach. After defining the initial domain O(xj)and the moving domainOðxjÞ, Eq. 6 after substitu-tion of Eq. 7 can be rewritten using the weakformulation in the deformed coordinate system as:Z

O

cti@ci@t

dO ¼ZO

�DID

Xj

cti;xj ci;xj þ ctiRxi

�dO

(

þZ@O

DIDctirci � nds ð11Þ

where ct is the test function, n unit normal vectorand ci;xj , the spatial derivatives of the concentra-tion of the species. The movement of the mesh(moving domain) (or swelling) was described bythe Poisson equation expressed in the weak formin the moving domain as:

ZO

utxj� uxjdO ¼ 0 ð12Þ

where u is the prescribed deformation obtainedfrom the swelling kinetics of the pellet resinategiven by:

u ¼ kc1 ð13Þ

where k is the swelling constant.Themapping between the original fixed domain

and thedeformingdomain canbederivedusing thechain rule, which results in the Jacobian, =:

= ¼ dxjdxj

ð14Þ

The mapping of the variables between the twodomains can be expressed as:

ci;xj ¼ =�1ci;xj ; cti;xj

¼ =�1cti;xj ;

uxj ¼ =�1uxj ; uti;xj

¼ =�1uti;xj

and dO

¼ detð=ÞdO

ð15Þ

where detð=Þ is the determinant of the Jacobian.The effect of the swelling on the diffusion

coefficients was described using a Fujita-typeexponential dependence27 expressed as:

Di ¼ Di;eq exp �bi 1� c1c1;eq

� �� ð16Þ

where bi and Di,eq are dimensionless constants

6 JEONG ET AL.


and diffusion coefficients at the equilibriumswollen state respectively.

Since the core matrix is not hydrated, it wasassumed that initially there was no free drug andno counter ions in the pellet. At the boundary(outer coating), the counter ion concentration wasconsidered as the equilibrium concentration and aperfect sink condition for the drug. These con-siderations result in the following initial andboundary conditions for the diffusing species.

ciðt ¼ 0; 8rÞ ¼ 0 ð17Þ

c1ðt > 0; r ¼ RÞ ¼ c1;eq ð18Þ

c2ðt > 0; r ¼ RÞ ¼ 0 ð19Þ

xjðt ¼ 0Þ ¼ xj ð20Þ

The amount of released drug was calculatedaccording to:

W ¼Zds

J � nds ð21Þ

where W is the average release rate and n, thenormal vector at the boundary. The amount ofreleased drug Mt is then given by:

Mt ¼Zdt

WðtÞdt ð22Þ

The fractional drug release M can then becalculated from:

F ¼ Mt

M1ð23Þ

where M1 is the amount of release after a longtime.

RESULTS AND DISCUSSION

Effect of Kollicoat1 SR 30D Coating Level on theDrug Release

The amount of DM loaded in 100 mg resincomplex was 65.19 mg (�0.51). The DM releaseprofiles of the coated uncoated resin particles areshown in Figure 2. The percentage level of coatingwas determined by the weight increase relative tothe amount of core. The drug release decreased asthe coating level increased and as a result thedrug release rate could bemodified easily by usingdifferent levels of the coating. Table 2 shows the

average particle size and coating thickness atdifferent coating levels.

When DM-loaded resin particles were coatedwith Kollicoat1 SR 30D, the release rates ofDM from the particles decreased significantlyeven with only 5% coating. This large decreasecan mainly be attributed to the large difference indiffusion coefficient of between the two materials.Due to the low cross-linking of the uncoated resinparticles, swelling takes place as they comein contact with aqueous environments such aswater or dissolution medium. This would result inincreased effective surface area and also change

Figure 2. In-vitro release of DM from different levelsof Kollicoat1SR30D coating using cores of DM/Dowex1

50WX4-200 resin particles in simulated gastric fluid(pH¼ 1.2) at 378C. The 5% coat indicates that theamount of coating is 5.0% w/w relative to the amountof core (drug/resin complex). Each point represent themean� standard error (n¼ 3).

Table 2. Average Particle Size of Coated andUncoated Ion-Exchange Resin Complexes and TheirCoating Thickness at Different Coating Levels

WeightGain(%)

Average CoatedParticle Diameter

(mm)

Estimated CoatingThickness

(mm)

0 106.2 (�27.7) —5 115.6 (�24.5) 4.710 122.7 (�17.3) 8.315 131.3 (�21.0) 12.620 138.3 (�25.1) 16.1



the effective diffusion coefficient of the resin.However, the presence of the coating constrainsthe swelling of the resin core which also contrib-uted to the observed difference.

The films ofKollicoat1SR30Dpolymerwithoutplasticizers were slightly brittle in dry state.However, when wet, they were flexible enough tobe elongated more than 100% so the crack forma-tion on the surface of coating due to the swelling ofthe core would be prevented. This was confirmedbymicroscopic observation discussed in the follow-ing section. It was observed that when 5% oftriethyl citrate (plasticizer)wasused theflexibilityof the polymer significantly increased, yielding anelongation value of more than 200%.

Microscopic Observation of the Coating

Scanning Electron Microscopy (SEM)

As shown in Figure 3A, only 5% coating coveredall areas of the resin particles without any cracksand uncoated areas. As the level of coatingincreased (Fig. 3B–D), the coat looked thickerbased on the surface morphology of the coatedparticles. The shape of the coated particles lookedlike a planet with many craters offering a non-uniformly coated surface. Several factors such asparticle-particle collisions during the fluid-bedcoating process, stresses that develop due to

gravitational forces and surface tension gradientsduring drying of the coating (Marangoni instabil-ity) might have contributed to the observednonuniformity. This affects the release profilesdue to the shorter or longer distances the drugmolecules travel through the non-uniformlycoated area in the films.

Confocal Laser Scanning Microscopy (CLSM)

The coated resin particles with different levels ofcoating were compared with respect to theirinternal structures using the CLSM, as shownin Figure 4A–D. In this experiment, the coatedpolymer membrane was visualized with Nile Red,a lipophilic fluorescence dye, and the coreappeared blue due to the self-fluorescence of theDM. The pictures show that with the coating levelincrease the intensity and thickness of the redcolor increased. However, there are many blackspots throughout the coating membrane showingthat the coating is not homogeneous and has somedefects. This confirms that there exist irregula-rities and anomalies in the coating, partiallyobserved in the SEM pictures.

These SEM pictures and CLSM images pro-vided valuable information for the developmentof the geometric model for the mathematicalsimulation in determining accurately the particlesize and film thickness. Moreover, the observed

Figure 3. Scanning electron micrographs of Kollicoat1 SR 30D coated DM-Dowex1

50WX4-200 resin particles, and the levels of coatings are 5% (A), 10% (B), 15% (C), and20% (D).

8 JEONG ET AL.


inhomogeneity and defects of the coating observedfrom these images helped diagnose the source ofpossible discrepancy between model and experi-mental data. This observation has lead to furthertheoretical investigation of the effects of theseanomalies on the drug release.

Drug Release Kinetics

Boyd Models

In Figure 5A and B, the DM release profiles fromthe uncoated pellets were entered into the particlediffusion (Fig. 5A) and film diffusion (Fig. 5B)controlled Boyd models. The linearity obtainedusing the particle diffusion controlled is higher(R2¼ 0.997) than that of the film diffusioncontrolled (R2¼ 0.904). This suggested that drugdiffusion in the resinate matrix of the uncoatedpellets was the rate-controlling step. The pre-sence of agitation can reduce the film layerthickness substantially decreasing the resistance

thus favoring the particle diffusion as ratecontrolling mechanism. It should be noted, how-ever, that due to the presence of ionic forces,strongly adhered ions on the surface of the pelletalways remain unaffected by the presence ofagitation providing a resistance layer. From theslope of the fitting curve, the diffusion coefficientof the resinate matrix was calculated and found tobe 3.4� 10�10 cm2/s (Tab. 3). This value, eventhough slightly higher, is in the same order ofmagnitude as the value reported in literature forthe same system (resinate)28 (1.24� 10�10 cm2/s).

In Figure 6A and B DM release profiles formcoated pellets (different coating levels) wereentered into the particle diffusion (Fig. 6A) andfilm diffusion (Fig. 6B) controlled Boyd models.Once more, for all the coating levels, the particlediffusion gives better linearity indicating that it isthe rate controlling mechanism as observed fromthe correlation coefficients. It should be notedhere that themodel does not differentiate betweenthe resin matrix and coating domains. Thus, the

Figure 4. Confocal laser scanning microscopic images of Kollicoat1 SR 30D coatedDM-Dowex1 50WX4-200 resin particles, and the levels of coatings are 5% (A), 10% (B),15% (C), and 20% (D). The polymer coating appears red due to the presence of Nile Red.The core appears blue due to the fluorescence of the drug. [Color figure can be seen in theonline version of this article, available on the website, www.interscience.wiley.com.]



diffusion coefficient calculated from the slopes ofthe coated pellets result in fictitious (lumped)value that is the resultant of the diffusivities of thecoating and the core polymers which, in thepresent case, differ by two orders of magnitude.It was observed that the slopes of the plots werestronglydependent on the coating level. Therefore,increasing the coating thickness by applying moreamount of polymer dispersion would be one way toreduce the diffusion rate through the coating.This, in turn, suggests that the particle diffusionthrough the coating polymer is particularly therate controlling mechanism. However, one of thelimitations is that one can not use thesemodels fordesign purposes with respect to the type of resin,

particle size, type and thickness of the coatingpolymer due to the fact that these domains (coreand coating) are treated as ahomogeneous system.

From the foregoing results, it is apparent thatthe film diffusion plays an insignificant role incontrolling the rate of release of DM from the ion-exchange pellets. However, to delineate theboundary between the two rate processes anddetermine the scope of the models with respect tothe environmental condition of the medium,further investigation was conducted. Boyd et al.22

argue that in the case where ionic concentrationof the ingoing ion in the release medium issmall, the film diffusion becomes the rate control-ling mechanism over the particle diffusion. In

Table 3. Model Parameters

Symbol/Unit Description Value

b1¼ 2¼ 3a Dimensionless Constant �0.31

D11 (cm2/s)b Diffusion Coefficient of Water in the Core 4.3� 10�6

D21 (cm2 s�1)b Diffusion Coefficient of Counter Ion in the Core 4.3� 10�6

D22 (cm2 s�1)c Diffusion Coefficient of Counter Ion in the Coating 1.0� 10�7

D31 (cm2 s�1)a Diffusion Coefficient of Drug in the Core (3.4� 1.1)� 10�10

D32 (cm2 s�1)c Diffusion Coefficient of Drug in the Coating (4.0� 1.3)� 10�12

c1,eq (g cm�3)d Equilibrium Concentration of Counter Ion 0.85M0 (g)

c Initial Drug Load 5.09� 10�7

aEstimated indirectly from measured data.bAdopted from Ref.29cMeasured.dFrom manufacturer manual.

Figure 5. DM release profiles from uncoated pellets plugged into particle diffusioncontrolled model (R2¼ 0.997) (A) and film diffusion controlled model (R2¼ 0.904) (B).

10 JEONG ET AL.


Figure 7, release profiles for the 20% w/w coatinglevel, where the ionic concentration in themediumwas reduced 10- and 100-fold, are depicted forparticle (Fig. 7A) and film (Fig. 7B) diffusioncontrolled models. In both cases, the particle

diffusion controlled process still gives a better fitto the linear curve than the film diffusion model.The fit quality was somewhat poor and this wasdue to the burst effect observed at the beginning ofthe release process. These ionic concentrations are

Figure 6. DM release profiles from coated pellets (different degree of coating) pluggedinto particle diffusion controlled model : 5% w/w coating (R2¼ 0.999); &: 10% w/wcoating (R2¼ 0.977); !: 15% w/w coating (R2¼ 0.993); ^: 20% w/w coating (R2¼ 0.999)(A); andfilmdiffusion controlledmodel : 5%w/wcoating (R2¼ 0.90);&: 10%w/wcoating(R2¼ 0.970);!: 15% w/w coating (R2¼ 0.92); ^: 20% w/w coating (R2¼ 0.88) (B).

Figure 7. DM release profiles from 20% w/w coated pellets at different ionicconcentration of the release medium plugged into particle diffusion controlled model( : 0.01 N HCl, R2¼ 0.942 and &: 0.001 N HCl, R2¼ 0.673) (A); and film diffusioncontrolled model (B) ( : 0.01 N HCl, R2¼ 0.743 and&: 0.001 N HCl, R2¼ 0.176).



far less than those encountered in a gastricenvironment. From all the foregoing findings, thefilm diffusion controlled process is insignificant inthe present enquiry.Hence, for the development ofthe mechanistic model it is plausible to consideronly the particle diffusion as a rate controllingmechanism.

Mechanistic Model

It is apparent that the use of the approximatemodels is limited to the identification of the ratecontrolling mechanism and do not help to conductfurther study with respect to other processes suchas swelling and those that are related to thedesign of the dosage forms. As a result, it isdeemed necessary to develop more detailedmodels for such applications. Swelling was con-sidered in the development of the detailednumerical model. Experimental swelling kineticsstudies show that 16 and 8% v/v swelling wasattained in equilibrium with the uncoated andcoated pellets, respectively. The swelling kineticswas extracted from the experimental results andwas used in the model to study the effect of theswelling on the DM release profile. Moreover,the dependence of the diffusion coefficient on theswelling was obtained by calculating the diffusion

coefficients from each measured time step of thedrug release from the uncoated pellets. Theobtained diffusion coefficients were fitted toEq. 16 and the constant b was found to be �0.31.The small value of b indicates that indeed theeffect of the swelling on the coefficient of diffusionis small. This constant was also used for thecounter ion and water diffusion dependence onswelling, with the measured values as thediffusivities at the equilibrium swollen state. InFigure 8A the comparison between simulatedswelling and experimental results is shown. It isworthwhile to emphasize that the pellets usedwere drug loaded and the resin when not loadedwith drug swells to about 33% v/v. The swellingkinetics of the pellets was well captured by themodel (Fig. 8A) as can be confirmed from the fitbetween the two curves visually. In Figure 8B,simulation results with and without the swellingas well as with the experimental data arecompared. To simulate the phenomenon whereswelling was ignored two diffusion coefficientvalues were used, namely the values before andafter swelling. It can be observed that (Fig. 8B)the swelling does not have substantial effect onthe drug release profile. Besides the small swel-ling level of the resinate complex, the swellingkinetics takes place faster (in the first 15–30 min

Figure 8. Comparison between simulated and experimental swelling kinetics for theuncoated pellets expressed as the percentage increase in radius. *: experiment; —:simulation (A); Comparison of simulation results with and without swelling as well aswith the experimental data for the uncoated pellets.*: experiment; : simulationwith swelling; : simulation with no swelling with diffusivity coefficient beforeswelling; : simulation with no swelling with diffusivity coefficient after swelling (B).

12 JEONG ET AL.


where only 20–30% of release takes place) ascompared to the diffusion process of the drug,which takes 3–4 h. It was observed that for thecoated pellets (results not shown) the swellingtakes place in the first couple of minutes andswells only to 8% v/v level. It should be noted thatthe inclusion of the swelling process in thedeveloped model substantially complicates themodeling building as well as solving process.Thus, in order to reduce the computational costand that the model remains useful for furtherstudies (such as, for instance, stochastic simula-tion of the coating nonuniformity) the swellingmodule can be shut off. In subsequent investiga-tion this approach was pursued.

Model Validation

Themodel was validated using experimental drugrelease profiles using different levels of coatingincluding 0% (Fig. 8B), 5, 10, 15, and 20% w/w.Comparisons between the experimental andsimulation results for the coated pellets areshown in Figure 9. The model has captured thedrug release dynamics of both the coated anduncoated pellets, as can be visually confirmed bythe fit between predictions and experimentalresults. To compare the observed fit quantita-tively, Root Mean Square Error (RMSE) valueswere calculated for each case. The calculated

RMSE ranges from 1.24 to 2.44% deviation. Themaximum deviation was observed for the 20%coated pellets. Due to the presence of a thickercoating, one expects an initial lag period whichthe model has predicted. However, the experi-mental results show what appears to be an initialburst phenomenon. This could be due to theexistence of a fraction of uncoated surface orunwashed drug particles on the surface of thecoating. This is apparent for the 5% coated pellets(Fig. 9). The SEM (Fig. 3) and confocal (Fig. 4)images show that presence of dents on the surfaceof the coated pellets. Defects and holes in thecoating were also observed in the confocal images.Thus, coating nonuniformity and variation inparticle size have played a role on the observeddiscrepancy.

It has already been shown that increasedpredictive accuracy was achieved by consideringthe dynamics description of all relevant processesinstead of quasi static treatment.12 The inclusionof the dynamic description of all the relevantprocesses and employing numerical approaches,an improvedmodel in terms of investigation of theeffect of processes usually ignored in the consid-eration of the simpler models was possible. Theother advantage of using numerical methods,especially Finite Element approach, is the flex-ibility in considering other geometries than theregular shapes thus enabling investigationof shape irregularities that can substantially

Figure 9. Simulated (solid lines) and experimental (symbols) profiles for &: 5% w/w;~: 10% w/w;*: 15% w/w; and }: 20% w/w coating.



influence the attainment of the desired releaseprofile during the design process.

CONCLUSIONS

Experimental and theoretical investigation ofcoated drug/ion-exchange complexes was pre-sented. From in vitro experiments it was revealedthat coating with as small as 5% w/w could affectthe release profile significantly, indicating diffu-sion of the drug from the particles via the coatingwas the rate determining step which was furtherconfirmed by the theoretical treatment usingapproximate analytical solutions of the Fick’sequation (the Boyd models). Diffusion of DMthrough the resin matrix was the rate-limitingmechanism in drug release from the uncoatedresin particles in contrast to the film diffusion.Further detailed mathematical model whichconsidered the different domains of the coatedparticles (the coating and core) and pertinentprocesses (ionic diffusion, swelling) was devel-oped. Swelling of the matrix for the uncoatedpellets had no significant effect on the drugrelease profile. The developed model has capturedthe drug release dynamics for the coated anduncoated ion-exchange pellets. The mathematicalmodeling can be used as a theoretical tool toinvestigate drug release form coated ion-exchangecomplexes for design purposes.

REFERENCES

1. Lowenthal W. 1972. Mechanism of action oftablet disintegrants: Correlation of tablet meanpore diameter and porosity. J Pharm Sci 61:303–304.

2. Elder DP. 2005. Pharmaceutical applications ofion-exchange resins. J Chem Educ 82:575–587.

3. Agarwal R, Mittal R, Singh A. 2000. Studies of ion-exchange resin complex of chloroquine phosphate.Drug Dev Ind Pharm 26:773–776.

4. Atyabi F, Sharma HL, Mohammad HAH, Fell JT.1996. Controlled drug release from coated floatingion exchange resin beads. J Control Release 42:25–28.

5. Bhaskar R, Murthy RSR, Miglani BD, Viswa-nathan K. 1986. Novel Method to evaluate diffusioncontrolled release of drug from resinate. Int JPharm 28:59–66.

6. Ichikawa H, Fujioka K, Adeyeye MC, Fukumori Y.2001. Use of ion-exchange resins to prepare 100mm-sized microcapsules with prolonged drug-

release by the Wurster process. Int J Pharm 216:67–76.

7. Langer DL. 1993. Polymer-controlled drug deliverysystems. Acc Chem Res 26:537–542.

8. Motycka S, Nairn JG. 1979. Preparation andevaluation of microencapsulated ion-exchangeresin beads. J Pharm Sci 68:211–215.

9. Borgquist P, Zackrisson G, Nilsson B, Axelsson A.2002. Simulation and parametric study of a film-coated controlled-release pharmaceutical. J Con-trol Release 80:229–245.

10. Farrell S, Sirkar KK. 2001. A mathematical modelof a hybrid disperssed network-membrane basedcontrolled release system. J Control Release 70:51–61.

11. Farrell S, Sirkar KK. 1999. A mathematical modelof an acqueous-organic partition based controlledrelease system using microporous membranes.J Control Release 61:345–360.

12. Frenning G, Tunon A, Alderborn G. 2003. Model-ling of drug release from coated granular pellets.J Control Release 92:113–123.

13. Koizumi T, Ritthidej GC, Phaechamud T.2001.Mechanistic modeling of drug release fromchitosan coated tablets. J Control Release 70:277–284.

14. Lee DJ, Liao YC. 1995. Slow-release from a coatedsphere with slightly deformed coating. J Pharm Sci84:1366–1373.

15. Liao YC, Lee DJ. 1997. Slow-release from a coatedsphere with slight deformations of coating film anddrug matrix. J Pharm Sci 86:92–100.

16. Sirotti C, Colombo M, Grassi M. 2002. Modelingof drug release from poly-disperse microencupsul-ated spherical particles. J Microencupsulation 19:603–614.

17. Du C, Zhou J, Shaviv A, Wang H. 2004. Mathema-tical model for potassium release from polymer-coated fertiliser. Biosyst Eng 88:395–400.

18. Lu SM, Lee SF. 1992. Slow release of Urea throughlatex film. J Control Release 18:171–180.

19. Higuchi T. 1961. Rate of release of medicamentsfrom ointment bases containing drugs in suspen-sion. J Pharm Sci 50:874–875.

20. Korsemeyer RW, Gurny R, Doelker EM, Buri P,Peppas NA. 1983. Mechanism of solute release-from porous hydrophilic polymers. Int J Pharm 15:25–30.

21. Peppas NA. 1985. Analysis of Fickian and non-Fickian drug release from polymers. Pharm ActaHelv 60:110–111.

22. Boyd GE, Adamson AW, Myers LS, Jr. 1947. Theexchange adsorption of ions from aqueous solutionsby organic zeolites. II. Kinetics. J Am Chem Soc 69:2836–2848.

23. Reichenberg D. 1953. Properties of ion-exchangeresins in relation to their structure. III. Kinetics ofexchange. J Am Chem Soc 75:589–597.

14 JEONG ET AL.


24. Costa P, Sousa Lobo JM. 2001. Modeling and compar-ison of dissolutionprofiles.EurJPharmSci 13:123–133.

25. Motycka S, Nairn JG. 1978. Influence ofwax coatings on release rate of anions from ion-exchange resin beads. J Pharm Sci 67:500–503.

26. Helfferich F. 1962. Ion Exchnage. editor, NewYork:McGraww-Hill. p 624.

27. Siepmann J, Peppas NA. 2001. Modeling of drugrelease from delivery systems based on hydroxy-

propyl methylcellulose (HPMC). Adv Drug DelivRev 48:139–157.

28. Pongjanyakul T, Prakongpan S, Rungsard-thong U, Chancham P, Priprem A. 2005.Characteristics and in vitro release of dextrome-thorphan resinates. Powder Technol 152:100–106.

29. Hering B, Bliss H. 1963. Diffusion in ion exchangeresins. AIChE J 9:495–503.



Documents

Drug Release Properties of Polymer Coated Ion-Exchange