Available online at www.sciencedirect.com
`
`SCIENCE @DIRECTO
`
`ELSEVIER
`
`International Journal of Pharmaceutics 269 (2004) 509-522
`
`international
`journal of
`pharmaceutics
`
`www.elsevier.com/locate/ijpharm
`
`Physicochemical properties and mechanism of drug release
`from ethyl cellulose matrix tablets prepared by
`direct compression and hot-melt extrusion
`Michael M. Crowley a,*, Britta Schroeder b, Anke Fredersdorfb, Sakae Obara c,
`Mark Talarico d, Shawn Kucera a, James W. McGinity a
`a Division of Pharmaceutics, College of Phannacy, The University of Texas at Austin, Austin, TX 78712, USA
`b College of Pharmacy, Freie Universitat Berlin, Berlin, Gennany
`c Shin-Etsu Chemical Co. Ltd., Specialty Chemicals Research Center, Niigata 942-8601, Japan
`d Micromeritics Instrument C01p., One Micromeritics Drive, Norcross, GA 30093-1877, USA
`Received 19 March 2003; received in revised fonn 16 September 2003; accepted 26 September 2003
`
`Abstract
`
`The objective of this research project was to determine the physicochemical properties and investigate the drug release
`mechanism from ethyl cellulose (EC) matrix tablets prepared by either direct compression or hot-melt extrusion (HME) of
`binary mixtures of water soluble drug (guaifenesin) and the polymer. Ethyl cellulose was separated into "fine" or "coarse"
`particle size fractions corresponding to 325-80 and 80-30 mesh particles, respectively. Tablets containing 30% guaifenesin were
`prepared at 10, 30, or 50 kN compaction forces and extruded at processing temperatures of 80-90 and 90-110 °C. The drug
`dissolution and release kinetics were determined and the tablet pore characteristics, tortuosity, thermal properties and surface
`morphologies were studied using helium pycnometry, mercury porosimetry, differential scanning calorimetry and scanning
`electron microscopy. The tortuosity was measured directly by a novel technique that allows for the calculation of diffusion
`coefficients in three experiments. The Higuchi diffusion model, Percolation Theory and Polymer Free Volume Theory were
`applied to the dissolution data to explain the release properties of drug from the matrix systems. The release rate was shown to
`be dependent on the ethyl cellulose particle size, compaction force and extrusion temperature.
`© 2003 Elsevier B.V. All rights reserved.
`
`Keywords: Hot-melt extrusion; Guaifenesin; Ethyl cellulose; Matrix tablets; Higuchi diffusion model; Porosity; Tortuosity; Percolation; Free
`volume; Sustained release; Mercury porosimetry
`
`1. Introduction
`
`Ethyl cellulose (EC) is a non-toxic, stable, com-
`pressible, inert, hydrophobic polymer that has been
`widely used to prepare pharmaceutical dosage forms.
`
`* Corresponding author. Tel.: + 1-512-471-6834;
`fax: +1-512-471-7474.
`E-mail address: mmcrowley@mail.utexas.edu (M.M. Crowley).
`
`The properties of ethyl cellulose sustained release
`products, including film coated tablets (Rowe, 1992),
`microspheres (Akbuga, 1991; Eldridge et al., 1990),
`microcapsules (Jalsenjak et al., 1977) and matrix
`tablets for both soluble and poorly soluble drugs
`(Shaikh et al., 1987a,b) have been reported.
`Hot-melt extrusion (HME) is a widely used pro-
`cess in the plastics industry to produce tubing, pipes
`and films. In pharmaceutical systems, HME has been
`
`0378-5173/$- see front matter© 2003 Elsevier B.V All rights reserved.
`doi: 10.1 016/j .ijphann.2003.09.037
`
`KASHIV1060
`IPR of Patent No. 9,492,393
`
`

`

`510
`
`MM Crowley eta!. /International Journal of Pharmaceutics 269 (2004) 509-522
`
`used to prepare granules, sustained-release tablets,
`and transdermal drug delivery systems (AitkenNichol
`et al., 1996; Follonier et al., 1994; McGinity et al.,
`2000; Zhang and McGinity, 1999, 2000). It has been
`demonstrated that the processing method can dictate
`the porosity and pore structure of the dosage form
`(Selkirk and Ganderton, 1970; Zoglio and Carstensen,
`1983). Previous workers have also shown that thermal
`processing results in a more tortuous product (Rubio
`and Ghaly, 1994; Zhang et al., 2001).
`Several researchers have suggested that diffusion
`controlled sustained release dosage forms prepared by
`HME have slower drug release rates than those pre-
`pared by traditional methods due to lower porosity and
`higher tortuosity (Young et al., 2002). Polymeric ma-
`terials are softened or molten during hot-melt extru-
`sion and subjected to intense mixing resulting in the
`generation of high pressures. Air present in the pow-
`der bed can be excluded from the polymer melt during
`hot-melt extrusion. As a result, HME dosage forms are
`expected to have a lower porosity and higher tortuos-
`ity, in comparison with the dosage forms prepared by
`tabletting processes.
`The Higuchi Square Root Model (Higuchi, 1963)
`has been successfully applied to model the kinetics
`of drug release from matrix systems. The Eq. (1) was
`derived from Fick' s Law of diffusion and applied to
`porous hydrophobic polymeric drug delivery systems
`in homogenous matrices and granular matrices.
`
`Q(t) = J D,C.~(2Co- sCa)t
`= J DappCa(2Co- cCa)t = k.Ji
`
`(1)
`
`In Eq. (1), Q(t) represents the cumulative amount of
`drug released at time t per unit surface area, Ds denotes
`the drug diffusion coefficient in the release medium,
`Co the total amount of drug in the matrix, Ca the sol-
`ubility of the drug in the release media, 8 the porosity,
`r the tortuosity of the matrix, Dapp the apparent or ob-
`served diffusion coefficient (Dapp = Dsc/r) and k the
`dissolution rate constant. Drug release can be manip-
`ulated by varying: (a) the initial concentration of drug
`within matrix; (b) porosity; (c) tortuosity; (d) the poly-
`mer system forming the matrix; and (e) the solubility
`of the drug.
`Drug release from a porous, hydrophobic polymeric
`drug delivery system occurs when the drug comes
`
`into contact with the release media, subsequently dis-
`solves and diffuses through media filled pores. Thus,
`the geometry and structure of the pore network are
`important in this process (Dees and Polderman, 1981;
`Lowenthal, 1972). The Higuchi model has been re-
`ported to fail at drug loading levels below the percola-
`tion threshold (Zhang and McGinity, 2000). Below the
`percolation threshold, incomplete drug release is ob-
`served presumably due to limited accessibility of many
`drug particles to the dissolution medium since they are
`encapsulated by water insoluble polymeric materials.
`The objectives of the present study were to deter-
`mine the physicochemical properties of ethyl cellulose
`matrix tablets prepared by either direct compression
`or hot-melt extrusion in order to explain the drug re-
`lease mechanism. The influence of compaction force
`and processing temperature on drug release rates and
`the physical properties of the tablets was studied. The
`effect of ethyl cellulose particle size was examined
`in tablets prepared by both direct compression and
`hot-melt extrusion. The median pore size, porosity and
`tortuosity of the matrix tablets prepared by the two
`techniques were determined using the Webb technique
`allowing calculation of the diffusion coefficients.
`
`2. Theory of mercury intrusion porosimetry and
`tortuosity
`
`Knowledge of tortuosity is necessary to determine
`diffusion coefficients from observed dissolution data.
`Researchers have relied upon secondary methods to
`determine tortuosity. These techniques required sev-
`eral steps and were time intensive. Recently, a novel
`approach using mercury intrusion porosimetry for di-
`rect measurement of tortuosity was reported (Webb,
`2001b). Mercury intrusion porosimetry has been used
`to study the pore characteristics of tablets (Gucluyildiz
`et al., 1977; Selkirk and Ganderton, 1970; Wikberg
`and Alderbom, 1992), granules (Opakunle and Spring,
`1976; Zoglio and Carstensen, 1983), ceramic particles
`for sustained drug delivery (Byrne and Deasy, 2002)
`and excipients (Riepma et al., 1993). The technique
`is based upon the unique properties of mercury. Mer-
`cury behaves as a non-wetting liquid toward most sub-
`stances and will not penetrate a solid unless pressure
`is applied. For circular pore openings, the Washburn
`Eq. (2) (Washburn, 1921) relates the applied pres-
`
`KASHIV1060
`IPR of Patent No. 9,492,393
`
`

`

`MM Crowley et a!. !International Journal of Pharmaceutics 269 (2004) 509-522
`
`511
`
`K =
`
`(3)
`
`r=
`
`(2)
`
`sure, P, and the radius, r, of the pores intruded with a
`non-wetting liquid:
`-2ycose
`p
`where y is the surface tension of mercury and e the
`contact angle between the liquid and sample. The in-
`verse relationship between pore radius and applied
`pressure indicates low pressures are used to measure
`large pore sizes and high pressures are used to mea-
`sure small pore sizes.
`Katz and Thompson (Katz and Thompson, 1986;
`Thompson et al., 1987) introduced an expression (3)
`for determination of permeability in porous rocks
`from mercury intrusion curves based upon con-
`cepts from Percolation Theory (Broadbent and Ham-
`mersley, 1957; Hammersley, 1957). These researchers
`found that absolute permeability, K, was related to
`the rock conductivity at a characteristic length Lc.
`Lmax
`¢S(Lmax)
`89 X Lc
`Eq. (3) requires dete1mination of the length at which
`conductance is at a maximum, Lmax, and the fraction
`of total porosity, ¢, filled at this length S(Lmax). The
`characteristic length, Lc, is the length at which mer-
`cury spans the entire sample and was found to be the
`point at which percolation begins. It is determined
`at the point of inflection in the rapidly rising region
`of the cumulative intrusion curve. Several different
`scientific disciplines have used the Katz-Thompson
`method to determine permeability in a variety of mate-
`rials (Bentz et al., 1998; Berkowitz and Balberg, 1993;
`Budd, 2002; Celzard et al., 2002; Garboczi and Bentz,
`2001).
`Jorgen Hager also derived an expression for mate-
`rial permeability based upon a capillary bundle model
`and knowledge of material tortuosity (Hager, 1998).
`The capillary bundle model describes the pore net-
`work as homogenously distributed in random direc-
`tions. Using the Hagen-Poiseuille correlation for fluid
`flow in cylindrical geometries in combination with
`Darcy's Law, Hager was able to derive an Eq. (4) for
`permeability, K, in terms of total pore volume, Vtot,
`material density, p, pore volume distribution by pore
`size, r~~~c.m_ax ry2 fv(ry) dry, and material tortuosity, T. In
`J 1}-1 c,rrun
`this method, Hager obtained all parameters except tor-
`tuosity from mercmy intrusion porosimetry analysis.
`
`K =
`
`1ry=rc,max
`1l fv(1J) dry
`
`(4)
`
`p
`24-r (1 + P Vtot)
`2
`lJ=rc min
`Webb concluded that combining the Hager and
`Katz-Thompson expressions (Eq. (5)) provided a
`means for determining tortuosity from mercury intru-
`sion porosimetry data (Webb, 2001a,b).
`
`r=
`
`11J=rc,max ryl fv(1J) dry
`lJ= rc,min
`
`(5)
`
`p
`24 K (1 + P Vfot)
`Since the determination of tortuosity depends upon
`permeability, changes in material permeability during
`the analysis will affect the reported value for tortuos-
`ity. Webb also reported that improved accuracy was
`achieved using true density data from helium pycnom-
`etry. Thus, the Webb technique allows determination
`of tortuosity from only two experiments, helium pyc-
`nometly and mercury intrusion. The apparent diffusion
`coefficient and diffusion coefficient can then be calcu-
`lated using the rate constant from dissolution studies.
`
`3. Experimental
`
`3.1. Materials
`
`Guaifenesin and glacial acetic acid were purchased
`from Spectrum Laboratory Products (Gardenia, CA).
`Ethyl cellulose grade Standard 10 (48.0--49.5%
`ethoxyl substitution, 9- 11 cP. viscosity range) was
`kindly donated by the Dow Chemical Company (Mid-
`land, MI). Methanol was purchased from EM Science
`(Gibbstown, NJ).
`
`3.2. Methods
`
`3.2.1. Preparation oftablets
`Ethyl cellulose was sieved into two fractions. The
`"coarse" fraction included the material that passed
`through a 30 mesh screen and was retained by an 80
`mesh screen. The "fine" fraction included material that
`passed through 80 mesh screen and was retained on
`a 325 mesh screen. Guaifenesin was passed through a
`30 mesh screen prior to use. The ethyl cellulose par-
`ticle size fractions were confirmed by introducing an
`aqueous suspension of ethyl cellulose to laser light
`scattering using a Malvern Mastersizer S (Malvern In-
`struments Limited, Malvern, Worcestershire, UK).
`
`KASHIV1060
`IPR of Patent No. 9,492,393
`
`

`

`512
`
`MM Crowley et al. /International Journal of Pharmaceutics 269 (2004) 509-522
`
`A model formulation containing 30% guaifenesin
`and 70% ethyl cellulose was selected for this study.
`The drug and polymer were geometrically diluted in
`a glass mortar and pestle and then introduced into
`a V Blender (Blend Master®, Patterson-Kelley, East
`Stroudsburg, PA) and mixed for 15 min. Tablets were
`prepared from the resultant blend by either direct com-
`pression or via hot-melt extrusion.
`
`3.2.2. Direct compression process
`Tablet compacts were prepared using a Carver
`25 Ton laboratory press (Fred Carver, Menomonee
`Falls, WI) and a 6 mm diameter flat faced punch and
`die set. The force applied on the punches was mea-
`sured using a load cell (ISI Inc., Round Rock, TX)
`mounted directly on the press with strain gauge sen-
`sors. Tablets weighing 250 ± 5 mg were compressed
`to 10, 30 or 50 kN for 3 s and ejected from the die.
`
`3.2.3. Hot-melt extrusion process
`The powder blend was fed into a single-screw Rand-
`castle Extruder (Model RC 0750, Cedar Grove, NJ)
`equipped with a Nitralloy 135M screw (3:1 compres-
`sion ratio with flight configuration containing feed,
`compression and mixing sections) and a rod shaped
`die (6 mm in diameter). The screw speed was 20 rpm.
`The three heating zones and die temperatures were set
`and allowed to equilibrate. Samples were prepared us-
`ing two different processing temperature ranges: "low"
`(80, 85, 85, 90 °C) and "high" (90, 105, 105, 110 °C).
`The residence time of the materials in the extruder was
`approximately 2-3 min. The extrudates were cooled
`to 45-55 oc and manually cut into tablets weighing
`250 ± 5mg.
`
`3.2.4. In vitro release properties
`Dissolution testing was performed using apparatus
`II on a Van Kel VK7000 Dissolution Tester (VanKel
`Industries, Edison, NJ) equipped with an auto sampler
`(Model VK 8000) according to the guaifenesin tablet
`monograph in USP 24. Six tablets were placed into the
`dissolution medium (900 ml of purified water) which
`was maintained at 37 oc by a circulating bath (Van
`Kel Model 750D) and agitated at 50 rpm. Samples
`(5 ml) were removed at specified time points over a
`12 h period without media replacement.
`Samples were analyzed for guaifenesin content us-
`ing a Waters (Milford, MA) high performance liquid
`
`chromatography (HPLC) system with a photodiode
`array detector (Model996) extracting at 276 run. Sam-
`ples were pre-filtered through a 0.45 1-Lm membrane
`(Gelman Laboratory, GHP Aero disc). An auto sampler
`(Model 717plus) was used to inject 20 1-11 samples. The
`data were collected and integrated using Empower®
`Version 5.0 software. The column was an Alltech
`Alltima™ cl8 3j.Lm, 150mm X 4.6mm. The mobile
`phase contained a mixture of water:methanol:glacial
`acetic acid in volume ratios of 600:400:15. The sol-
`vents were vacuum filtered through a 0.451-Lm nylon
`membrane and degassed by sonication. The flow rate
`was 1.0 mVmin. The retention time of the guaifen-
`esin was 4 min. Linearity was demonstrated from 2
`to 200 j..Lg/ml (R2 ~ 0.997) and injection repeata-
`bility was 0.35% relative standard deviation for 10
`injections.
`
`3.2.5. True density by helium pycnometry
`The true density of the powder formulations and
`tablets was determined in triplicate using helium pyc-
`nometry (Micromeritics ® AccuPyc 1330 Pycnometer;
`Norcross, GA). Twelve tablets were placed in a 12 cm3
`sample cup and purged 20 times at 19.85 psi followed
`by six analytical runs at 19.85 psi. The equilibration
`rate was 0.0050 psi/min. An equivalent mass of the
`powder mixtures was measured in the same manner.
`
`3.2. 6. Mercury intrusion porosimetry
`The bulk density and tortuosity of the tablets were
`determined using an AutoPore IV 9500 mercury intru-
`sion porosimeter (Micromeritics, Norcross, GA). In-
`cremental volumes of mercury were plotted against
`pore diameters according to the Washburn Eq. (2). A
`surface tension value of 485 Diem was used for mer-
`cury and its contact angle was 130°. Twelve tablets
`were placed in a Sec bulb penetrometer and pressure
`was applied from 1 to 15,000 psia, representing pore
`diameters of0.012-360 j..Lm. The pressure at each point
`was allowed to equilibrate for 10 s. Each run was per-
`formed in triplicate. Tablets were analyzed prior to and
`post-dissolution, with the removal of water from the
`tablets following dissolution by drying to a constant
`weight under vacuum for a minimum of 72 h.
`
`3.2. 7. Percent effective porosity determination
`Percent effective porosity, s was determined ac-
`cording to the Varner technique (Varner, 1991) us-
`
`KASHIV1060
`IPR of Patent No. 9,492,393
`
`

`

`MM Crowley et al. I International Journal of Pharmaceutics 269 (2004) 509-522
`
`513
`
`ing Eq. ( 6) in which Pt is the true density of the
`tablet as determined by helium pycnometry and Ph
`the bulk density as determined by mercury intrusion
`porosimetry. Three runs of 12 tablets were analyzed
`first by helium pycnometry followed by mercury in-
`trusion porosimetry.
`
`of Moore and Planner for dissolution curve compari-
`son using the similarity (fi) and difference factors (/2)
`(Moore and Planner, 1996).
`
`4. Results and discussion
`
`£=
`
`Pt- Pb
`Pb
`
`(6)
`
`4.1. The influence of ethyl cellulose particle size on
`drug release
`
`3.2. 8. Modulated differential scanning calorimetry
`analysis
`scanning
`Temperature modulated differential
`calorimetry (M-DSC) was used to characterize the
`thermal properties of the polymer and drug in phys-
`ical mixtures and hot-melt extrudates. M-DSC anal-
`ysis was conducted using a Thermal Advantage
`Model 2920 from TA Instruments (New Castle, DE)
`equipped with Universal Analysis 2000 software. Ul-
`trahigh purity nitrogen was used as the purge gas at a
`flow rate of 150 mVmin. The sample was weighed to
`10 ± 5 mg and placed in aluminum pans (Kit 0219-
`0041, Perkin-Elmer Instruments, Norwalk, CT) and
`crimped with an aluminum lid. The temperature ramp
`speed was 2 °C/min from 25 to 150 oc for all studies
`with a modulation rate of 1.592 oc every minute.
`
`3.2.9. Scanning electron microscopy
`Scanning electron microscopy was used to study the
`surface morphology of the hot-melt extruded tablets.
`The samples were mounted on an aluminum stage us-
`ing adhesive carbon tape and placed in a low humid-
`ity chamber for 12 h prior to analysis. Samples were
`coated with gold-palladium for 60 s under an argon
`atmosphere using a Pelco® Model 3 Sputter Coater
`(Ted Pella Inc., Tusin, CA) in a high vacuum evapora-
`tor equipped with an omni-rotary stage tray. Scanning
`electron microscopy was performed using a Hitachi
`S-4500 field emission microscope operating at an ac-
`celerating voltage of 15 kV and a 15 J.LA emission cur-
`rent. Images were captured with Quartz® software.
`
`3.2.1 0. Statistical analysis
`One-way analysis of variance (ANOVA) was used to
`determine statistically significant differences between
`results. Results with P-values <0.05 were considered
`statistically significant (a= 0.05). Dissolution curves
`were analyzed by the model independent approach
`
`The influence of ethyl cellulose particle size and
`processing conditions on the release rate of guaifen-
`esin from matrix tablets was studied. The influence
`of ethyl cellulose particle size, compaction force and
`extrusion temp.erature on guaifenesin release rate
`is presented in Fig. 1. The guaifenesin release rate
`from the matrix was highly dependent upon ethyl
`cellulose particle size and the processing conditions
`employed to prepare the tablet. In both cases, slower
`release rates were observed with small particle size
`ethyl cellulose.
`According to Percolation Theory, when a matrix is
`composed of a water soluble drug and a water insolu-
`ble polymer, drug release occurs by dissolution of the
`active ingredient through capillaries composed of in-
`terconnecting drug particle clusters and the pore net-
`work (Holman and Leuenberger, 1988; Leuenberger
`et al., 1987). As drug release continues, the intercon-
`necting clusters increase the pore network through
`which interior drug clusters can diffuse. The total
`number of ethyl cellulose particles increases when its
`pmiicle size is reduced. With more ethyl cellulose
`particles present, the theory predicts that fewer clus-
`ters of soluble drug substance are formed. Further-
`more, the presence of finite drug clusters ( encap-
`sulated drug particles) is more statistically plausi-
`ble. The resulting pore network becomes less ex-
`tensive and more tortuous resulting in slower drug
`release.
`Mercury porosimetry and helium pycnometry were
`used to determine the pore characteristics and tor-
`tuosity of the tablets prior to dissolution testing.
`The median pore radius, percent effective porosity
`and tortuosity of the matrix tablets are presented in
`Table 1. Statistically significant differences in median
`pore radius (n = 3, P < 0.006), porosity (n = 3,
`P < 0.02) and tortuosity (n = 3, P < 0.009) were
`observed between matrix tablets prepared with "fine"
`
`KASHIV1060
`IPR of Patent No. 9,492,393
`
`

`

`514
`
`MM Crowley et al. /International Journal of Pharmaceutics 269 (2004) 509-522
`
`'0
`Cl)
`
`II> m 60
`Cl)
`-a;
`
`a:::: -c: Cl)
`
`(.)
`r...
`Cl)
`D..
`
`80
`
`40
`
`20
`
`0
`
`2
`
`{A)
`
`4
`8
`6
`Time (Hours)
`
`10
`
`12
`
`0
`
`2
`
`(B)
`
`4
`
`8
`6
`Time (Hours)
`
`10
`
`12
`
`Fig. 1. Influence of ethyl cellulose particle size, compaction force
`and extmsion temperature on guaifenesin release from matrix
`tablets prepared by direct compression and hot-melt extmsion
`containing 30% guaifenesin and 70% ethyl cellulose using USP
`Method II at 3 7 o C and 50 rpm in 900 ml of purified water. Each
`point represents the mean± standard deviation, n = 6. (A) Matrix
`tablets prepared using "fine" ethyl cellulose (325-80 mesh) and
`(B) matrix tablets prepared using "coarse" ethyl cellulose (80--30
`mesh). (+) Direct compression, 10 kN; (.) direct compression,
`30 kN; ( ... ) direct compression, 50 kN; c•) hot-melt extmsion, 80,
`85, 85, 90 oc; (D) hot-melt extmsion, 90, 105, 105, 11 o oc.
`
`and "coarse" ethyl cellulose particle sizes. The me-
`dian pore radius was smaller, the tablets less porous
`and more tortuous when prepared with "fine" ethyl
`cellulose (137-529 A, 0.4-4.1% porosity and 4.7-321
`tortuosity) than in tablets prepared with "coarse"
`ethyl cellulose (351-1439 A, 0.7-6.5% porosity and
`1.3-125 tortuosity). These data are supported by the
`dissolution results.
`Katikaneni and coworkers (Katikaneni et al.,
`1995a,b; Upadrashta et al., 1993, 1994) reported that
`pseudoephedrine hydrochloride release from ethyl
`cellulose matrix tablets prepared by direct compres-
`sion was controlled by ethyl cellulose particle size
`and compression force. These authors concluded that
`the drug release rate decreased due to a reduction
`in porosity and increased matrix tortuosity at high
`compaction forces or when a finer particle of size
`ethyl cellulose was used. The findings in the present
`study confirm the work of Katikaneni, support the
`predictions of Percolation Theory and demonstrate
`the utility of the mercury porosimetry technique for
`determination of tortuosity.
`Generally, pore radii greater than 200 A are neces-
`sary to avoid hindered diffusion (Anderson and Quinn,
`1974). Hindered diffusion can be due to steric exclu-
`sion or hindered particle motion. In these situations,
`either the pore radii are sufficiently small that the
`molecular dimensions of the solute restrict its diffu-
`sion or excessive frictional resistance is created. Hin-
`dered diffusion is typically observed when values of
`the particle radius to pore wall radius ratio are greater
`than 0.1. The molecular dimensions of guaifenesin
`were modeled using SAVOL3 software to assess the
`likelihood of hindered diffusion (Pearlman and Skell,
`2003). Assuming a spherical shape, guaifenesin was
`found to have a radius of 3.53 A excluding a solvent
`radius and 5.44 A with a solvent radius. Assuming
`an ovoid shape and including a solvent radius, the
`model predicted the major axis to be 7. 84 A, the first
`minor axis to be 6.13 A and the second minor axis
`to be 3.89 A. Thus, the minimum guaifenesin particle
`radius to pore radius ratio is 0.04 in the case of the
`hot-melt extruded tablets prepared using "fine" ethyl
`cellulose and processed at 90-110 °C. This value
`is sufficiently small that hindered diffusion can be
`considered negligible.
`Several assumptions are made using the mercury
`porosimetry technique which can introduce consid-
`
`KASHIV1060
`IPR of Patent No. 9,492,393
`
`

`

`MM Crowley et al. / International Journal of Pharmaceutics 269 (2004) 509-522
`
`515
`
`Table 1
`Median pore radius, percent porosity and tortuosity of matrix tablets prepared by direct compression (DC) and hot-melt extrusion (HME)
`containing 30% guaifenesin and 70% ethyl cellulose measured before and after dissolution testing
`
`Processing conditions
`
`Median pore radius (A)
`
`Before
`
`After
`
`Percent porosity
`
`Before
`
`After
`
`Tortuosity
`
`Before
`
`4.7±1.1
`6.0 ± 0.6
`6.9 ± 0.9
`53.2 ± 8.1
`321 ± 22
`
`1.3 ± 0.7
`2.1 ± 0.6
`2.8 ± 0.5
`41.8 ± 5.5
`125 ± 16
`
`After
`
`4.7 ± 0.6
`5.2 ± 0.9
`
`4.4 ± 0.3
`5.0 ± 1.1
`
`4.1 ± 0.3
`3.8 ± 0.2
`3.1 ± 0.2
`1.0 ± 0.2
`0.4 ± 0.2
`
`6.5 ± 0.2
`5.7 ± 0.3
`5.2 ± 0.2
`2.3 ± 0.4
`0.7 ± 0.1
`
`8.1 ± 1.8
`7.8 ± 2.2
`
`8.7 ± 1.4
`8.0 ± 1.1
`
`"Fine" ethyl cellulose (325-80 mesh)
`DC lOkN
`529 ± 24
`DC 30kN
`476 ± 26
`DC 50kN
`475 ± 17
`HME 80, 85, 85, 90 ac
`264± 8
`HME 90, 105, 105, 110 ac
`137±6
`"Coarse" ethyl cellulose (80-30 mesh)
`DC 10kN
`1439 ± 42
`DC 30kN
`906 ± 23
`DC 50kN
`710 ± 25
`HME 80, 85, 85, 90 ac
`642 ±48
`392 ± 11
`HME 90, 105, 105, 110 ac
`588 ±57
`351 ± 13
`Each point represents the mean± standard deviation, n = 3.
`
`575 ±51
`521 ± 65
`
`erable error into the measurements. The assumption
`of circular pore cross-sections or "cylindrical pore
`geometry" is due to mathematical convenience in
`order to avoid complexities calculating mean radii
`and contact angles in irregularly shaped pores. As
`a result, the technique is biased to calculate smaller
`pore sizes than are actually present. Furthermore,
`"ink bottle pores" also bias the pore size calculation
`to the small side. Ink bottle pores are defined as
`pores in which the diameter increases as one moves
`to its inner dimensions (Leon, 1998; Salmas and
`Androutsopoulos, 2001). Mercury intrusion also in-
`volves subjecting the tablets to equal pressures from
`360° as the mercury is forced inside. This means
`that as mercury intrudes into the pores, the walls of
`pores penetrated by mercury at a given pressure are
`uniformly affected by the same pressure. This makes
`collapse of a pore wall unlikely. However, the tablet
`can compress under the applied pressure which allows
`for additional mercury to intrude. This also biases
`measurement to calculation of smaller than actual
`pore size. Obviously, the compression of ethyl cellu-
`lose is a concern using this technique. To minimize
`the impact of tablet compression, the applied pressure
`was not allowed to exceed that applied during the di-
`rect compression experiments (50 kN or 15,000 psia).
`Despite these limitations, pore size information ob-
`tained from mercury porosimetry experiments has
`been consistently demonstrated to be representative.
`
`4.2. The Influence of tableting technique on drug
`release
`
`Guaifenesin release rates were considerably slower
`in tablets prepared by hot-melt extrusion than those
`prepared by direct compression as shown in Fig. I.
`Statistically significant differences in median pore
`radius (n = 3, P < 0.006), percent effective poros-
`ity (n = 3, P < 0.002) and tortuosity (n = 3,
`P < 0.0004) were observed between the direct com-
`pression and hot-melt extruded tablets as shown in
`Table 1. The median pore radius was smaller and the
`tablets less porous when prepared by hot-melt extru-
`sion (137-392 A, 0.4-2.3% porosity and 41.8-321
`tortuosity) than by direct compression (475-1439 A,
`3.1-6.5% porosity and 1.3-6.9 tortuosity). Adsorption
`and binding studies did not reveal any interactions.
`Complete drug recovery was obtained in samples
`dissolved in ethanol. Thus, the differences in release
`rate are due to the bonding mechanism of the parti-
`cles in the two tableting techniques. Ethyl cellulose
`has been reported to be ductile and the predominant
`mechanism during compaction is plastic deformation
`(Katikaneni et al., 1995b). During hot-melt extrusion,
`solid bridges are formed upon cooling.
`The extrusion temperatures were above the melt-
`ing point of guaifenesin but below the glass transition
`temperature of ethyl cellulose. The molten drug can
`act as a solvent for the polymeric particles. Polymeric
`
`KASHIV1060
`IPR of Patent No. 9,492,393
`
`

`

`516
`
`MM Crowley et a!. I International Journal of Pharmaceutics 269 (2004) 509-522
`
`particles that do not go into solution are softened at
`elevated temperatures and deformed by the rotating
`screw during extrusion. Upon cooling, extensive solid
`bridges are formed between the particles resulting in a
`smaller pore radius and more tortuous network. These
`results demonstrate that densification of the molten
`mass during extrusion occurs to a greater extent rela-
`tive to compaction of the powder during compression.
`Thermal treatment of amorphous polymers has
`also been shown to decrease polymer free volume
`(Follonier et al., 1995). Elevated temperatures and
`high pressures during hot-melt extrusion followed by
`cooling significantly impact the free volume. Polymer
`chain motion and free volume increase with temper-
`ature allowing molten guaifenesin molecules to enter
`these voids. Upon cooling, the guaifenesin molecules
`
`remain dispersed in these domains. The net effect is
`an increase in the degree of packing and a reduc-
`tion in free volume. Diffusion of molecules through
`amorphous polymers is governed by free volume and
`packing. Thus, a decrease in drug transport is antic-
`ipated when the free volume decreases and packing
`order increases.
`The miscibility of guaifenesin in ethyl cellulose was
`studied by modulated differential scanning calorime-
`try. The glass transition temperature of ethyl cellulose
`was found to be 129.3 oc (Fig. 2A) and the melting
`point of guaifenesin was found to be 85.6 oc (Fig. 2B).
`The melting point of guaifenesin can be observed in
`the thermograms of the physical mixtures (Fig. 2C
`and F). Thermo grams of the hot-melt extrudates pro-
`cessed at 80-90 °C show a melt transition about 9 oc
`
`3.750
`
`3.125
`
`2.500
`
`1.875
`
`1.250
`
`...........
`C)
`
`0.625
`~ 0.000
`~ -0.625
`u::
`1ii
`(I)
`I
`
`-1.250
`
`-1.875
`
`-2.500
`
`-3.125
`
`-3.750
`
`-4.375
`
`-5.000
`20
`
`ExoUp
`
`A
`
`B
`
`c
`
`129.27"C(I)
`
`85.64°C
`
`D ~
`
`aa.saoc
`~---------------------~
`
`E
`
`77.52°C
`
`F
`
`86.73°C
`
`75.16°C
`
`40
`
`60
`
`100
`80
`Temperature ("C)
`
`120
`
`160
`140
`Universal V3.0G TA lnstrum ents
`
`Fig. 2. Modulated differential scanning calorimetry profiles of ethyl cellulose, guaifenesin, physical mixtures and hot-melt extrudates. (A)
`Ethyl cellulose; (B) guaifenesin; (C) physical mixture of 30% guaifenesin and 70% "coarse" ethyl cellulose; (D) hot-melt extrudate of 30%
`guaifenesin and 70% "coarse" ethyl cellulose processed at 80, 85, 85, 90 °C; (E) hot-melt extrudate of 30% guaifenesin and 70% "coarse"
`ethyl cellulose processed at 90, 105, 105, l10 °C; (F) physical mixture of 30% guaifenesin and 70% "fine" ethyl cellulose; (G) hot-melt
`extrudate of 30% guaifenesin and 70% "fine" ethyl cellulose processed at 80, 85, 85, 90 oc; (H) hot-melt extrudate of 30% guaifenesin
`and 70% "fine" ethyl cellulose processed at 90, 105, 105, 110 °C.
`
`KASHIV1060
`IPR of Patent No. 9,492,393
`
`

`

`below the melting point of guaifenesin (Fig. 2D and
`G). Thermal transitions corresponding to the melting
`point of guaifenesin are not observed in the extrudates
`processed at 90-100 oc (Fig. 2E and H). These results
`indicate that guaifenesin was miscible with ethyl cel-
`lulose while in the molten state. The presence of a tran-
`sition in the hot-melt extrudates processed at 80-90 oc
`suggests that a two phase solid dispersion was formed
`in which guaifenesin was present in both amorphous
`and crystalline fmm. The absence of a transition in the
`extrudates processed at high