Journal of Controlled Release 82 (2002) 137–147
`
`www.elsevier.com/locate /jconrel
`
`P recise control of PLG microsphere size provides enhanced
`control of drug release rate
`
`a
`a
`a
`b
`Cory Berkland , Martin King , Amanda Cox , Kyekyoon (Kevin) Kim ,
`a ,*
`Daniel W. Pack
`aDepartment of Chemical Engineering,University of Illinois, MC-712, Box C-3,600 South Mathews Avenue,Urbana, IL 61801,USA
`bDepartment of Electrical and Computer Engineering,University of Illinois,Urbana, IL 61801,USA
`
`Received 6 April 2002; accepted 17 May 2002
`
`Abstract
`
`An important limitation in the development of biodegradable polymer microspheres for controlled-release drug delivery
`applications has been the difficulty of specifically designing systems exhibiting precisely controlled release rates. Because
`microparticle size is a primary determinant of drug release, we developed a methodology for controlling release kinetics
`employing monodisperse poly(D,L-lactide-co-glycolide) (PLG) microspheres. We fabricated 20-, 40- and 65-mm diameter
`rhodamine-containing microspheres and 10-, 50- and 100-mm diameter piroxicam-containing microspheres at various
`loadings from 1 to 20%. In vitro release kinetics were determined for each preparation. Drug release depended strongly on
`microsphere diameter with 10- and 20-mm particles exhibiting concave-downward release profiles while larger particles
`resulted in sigmoidal release profiles. Overall, the rate of release decreased and the duration increased with increasing
`microsphere size. Release kinetics from mixtures of uniform microspheres corresponded to mass-weighted averages of the
`individual microsphere release kinetics. Appropriate mixtures of uniform microspheres were identified that provided constant
`(zero-order) release of rhodamine and piroxicam for 8 and 14 days, respectively. Mixing of uniform microspheres, as well as
`control of microsphere size distribution, may provide an improved methodology to tailor small-molecule drug-release
`kinetics from simple, biodegradable-polymer microparticles.  2002 Elsevier Science B.V. All rights reserved.
`
`Keywords: Controlled release; Zero-order release; Uniform microspheres; Poly(lactide-co-glycolide); Piroxicam
`
`1 . Introduction
`
`In comparison to conventional dosage forms,
`biodegradable polymeric matrices provide improved
`delivery methods for small molecules, peptides,
`proteins and nucleic acids. By encapsulating the drug
`
`*Corresponding author. Tel.: 11-217-244-2816; fax: 11-217-
`333-5052.
`E-mail address: dpack@uiuc.edu (D.W. Pack).
`
`in a polymer matrix from which it is released at a
`relatively slow rate over a prolonged time, controlled
`release affords less frequent administration, thereby
`increasing patient compliance and reducing discom-
`fort; protection of the therapeutic compound within
`the body; potentially optimized therapeutic responses
`and prolonged efficacy; and avoidance of peak-re-
`lated side-effects by maintaining more-constant
`blood levels of the drug. Further, because such
`devices can be administered by injection, one can
`
`0168-3659/02/$ – see front matter  2002 Elsevier Science B.V. All rights reserved.
`PII: S0168-3659( 02 )00136-0
`
`ALKERMES EXHIBIT 2035
`Amneal Pharmaceuticals LLC v. Alkermes Pharma Ireland Limited
`IPR2018-00943
`
`Page 1 of 11
`
`

`

`138
`
`C. Berkland et al. / Journal of Controlled Release 82(2002)137–147
`
`also achieve localized drug delivery and high local
`concentrations.
`The large and growing variety of pharmaceuticals
`on the market and in development require versatile
`delivery systems that can adapt
`to the needs of
`particular applications [1], especially the capacity to
`generate the required delivery rates and, perhaps,
`variation of delivery rate over time. For example,
`many therapeutics require a constant release rate for
`varying durations from several days to several weeks
`[2–6]. Such ‘zero-order’ release is a long-sought
`goal of controlled-release drug delivery, but has been
`difficult
`to achieve for many pharmaceuticals. In
`contrast, variable drug release rates can be beneficial
`for many important indications [7]. Intermittent high
`doses of antibiotics may alleviate evolution of resist-
`ance in bacteria, and discontinuous administration of
`vaccines often enhances
`the immune
`response
`[2,8,9].
`Microparticle drug delivery systems may provide
`the needed versatility. Drug release rates can be
`controlled through the choice of polymer chemistry
`[10,11]
`(e.g. polymer composition, co-monomer
`ratios, molecular weight, etc.) or variation of the
`microparticle formulation parameters, and thus the
`physical characteristics of
`the resulting particles
`[12,13]. Nevertheless,
`the ability to tailor drug
`release kinetics is limited. For typical small-molecule
`therapeutics, as well as some proteins [11,14–17],
`drug release often exhibits an initial ‘burst’ phase
`during which a significant fraction (typically 5–
`50%) of the encapsulated compound is released in a
`short time (,24 h). The burst is usually undesirable
`because the drug that is released in this phase is not
`available for prolonged release, and more important-
`ly for potent therapeutics or drugs with a narrow
`therapeutic window, this initial bolus may result in
`toxicity or other side-effects. The burst may be
`followed by a lag phase exhibiting negligible release
`and, more typically, a phase in which the release rate
`decreases with time due to a decreasing driving force
`as drug is depleted from the matrix. Various strate-
`gies for reducing or eliminating the initial burst have
`been studied including chemistry (block copolymers
`with hydrophilic regions) [10], variation of micro-
`sphere formation parameters [12,13], coating of
`microspheres (microencapsulated microspheres) [18]
`and conjugation of drug to the polymer matrix [19].
`Microsphere size is a primary determinant of drug
`
`release rates. Larger spheres generally release en-
`capsulated compounds more slowly and over longer
`time periods, other properties (polymer molecular
`weight, initial porosity, drug distribution within the
`sphere, etc.) being equal. Thus, controlling sphere
`size provides an opportunity for control of release
`kinetics. Numerous studies have been conducted to
`determine the effects of sphere size on drug release
`[3,10,11,13,20,21]. However, due to a limited ability
`to control microsphere size, this approach to modu-
`lating release rates has been relatively unexplored.
`We have devised a methodology for precisely
`controlling microsphere size and size distribution
`[20]. Our spraying technology is capable of generat-
`ing uniform PLG microspheres ranging in size from
`about 1 to .500 mm. For example, we recently
`reported fabrication of microspheres with diameters
`of |5–80 mm, wherein 95% of the particles had a
`diameter within 1.0–1.5 mm of the average [20].
`Furthermore, the methodology allows fabrication of
`novel, continuously varying size distributions of any
`desired shape. We hypothesized that the ability to
`control particle size afforded by our system would
`lead to enhanced control of drug release kinetics.
`Here we report release kinetics for a model com-
`pound, rhodamine B, and the non-steroidal anti-
`inflammatory drug (NSAID) piroxicam from uni-
`form PLG microspheres. While piroxicam is similar
`in molecular weight to rhodamine, these two com-
`pounds were chosen to represent water-soluble
`(rhodamine, 7.8 mg/ml) and -insoluble (piroxicam,
`53.3 mg/ml at pH|7) drugs [22]. We demonstrate
`that
`the release kinetics of both compounds are
`indeed variable depending on the microsphere size,
`as expected. Further, we show that mixtures of
`uniform microspheres exhibit release kinetics that are
`weighted averages of
`the individual microsphere
`release kinetics. Based on this finding, we chose
`appropriate mixtures to generate zero-order release,
`without an initial burst phase, for both rhodamine
`and piroxicam.
`
`2 . Materials and methods
`
`2 .1. Materials
`
`lactic
`(50:50
`Poly(D,L-lactide-co-glycolide)
`acid:glycolic acid; i.v.50.20–24 dl/g corresponding
`
`Page 2 of 11
`
`

`

`C. Berkland et al. / Journal of Controlled Release 82(2002)137–147
`
`139
`
`to M 10,000–15,000) was obtained from Birming-
`w
`ham Polymers. Poly(vinyl alcohol) (PVA; 88% hy-
`drolyzed) was
`obtained
`from Polysciences.
`Rhodamine B chloride was obtained from Sigma.
`Piroxicam free base was a gift
`from Dongwha
`Pharmaceuticals (Seoul, Korea). HPLC grade di-
`chloromethane (DCM), dimethylsulfoxide and so-
`dium hydroxide were purchased from Fisher Sci-
`entific.
`
`2 .2. Preparation of microspheres
`
`Microspheres were prepared as described previ-
`ously [20]. Briefly, PLG solutions (5% (w/ v) in
`DCM) containing rhodamine B or piroxicam at the
`various
`concentrations
`indicated were pumped
`through a small glass nozzle at various flow rates,
`while an ultrasonic transducer (Branson Ultrasonics)
`controlled by a frequency generator (Hewlett Pac-
`kard model 3325A) disrupted the stream into uni-
`form droplets. A carrier stream (1% (w/ v) PVA in
`distilled water) flowed around the emerging PLG
`stream. The streams flowed into a beaker containing
`|500 ml of 1% PVA, and the particles were stirred at
`room temperature for 3 h, filtered, and rinsed with
`distilled water. The microspheres were lyophilized
`(Labconco benchtop model) for a minimum of 48 h
`and were stored at 220 8C under desiccant.
`
`2 .3. Determination of drug loading
`
`loading of rhodamine B was deter-
`The initial
`mined as follows. A known mass (|2–5 mg) of
`microspheres was dissolved in 50 ml dimethylsulfox-
`ide. PBS (500 ml) was added and precipitated
`polymer was removed by centrifugation at 12,000
`rpm for 10 min. Rhodamine B concentration in the
`supernatant was determined by measuring the ab-
`sorbance at 550 nm in a multi-well plate spec-
`trophotometer
`(Molecular Devices Spectra Max
`340PC).
`To determine piroxicam loading, a known mass
`(|5 mg) of microspheres containing piroxicam was
`dissolved in 1 ml of 0.25 M sodium hydroxide at
`room temperature for 5 min. Blank (piroxicam free)
`microspheres of the same size were treated identical-
`ly. Piroxicam concentration in the resulting solution
`was determined by measuring the absorbance at 276
`
`nm (Varian Cary 50) in a quartz cuvette and subtract-
`ing absorbance values for the blank microspheres.
`
`2 .4. In vitro drug release
`
`Rhodamine release was determined by resuspend-
`ing a known mass of microspheres encapsulating
`rhodamine B in 2 ml of phosphate-buffered saline
`(PBS, pH 7.4) containing 0.5% Tween. The suspen-
`sions were continuously agitated by inversion (at
`|10 rpm) in a 37 8C incubator. At regular intervals
`the samples were centrifuged, the supernatant was
`removed, and the spheres were resuspended in fresh
`PBS. Concentration of rhodamine B in the superna-
`tant was determined using the spectrophotometer as
`described above. The amount of rhodamine in each
`sample was summed with the amounts at each
`previous time point, and the total divided by the
`amount of
`rhodamine in the microspheres (ex-
`perimental loading3mass of microspheres), to arrive
`at the ‘cumulative percent released’.
`Piroxicam release was determined by resuspending
`|5 mg of microspheres in 1.3 ml of PBS containing
`0.5% Tween. Conditions during drug release were
`the same as described above for rhodamine. After
`centrifugation, the concentration of piroxicam in the
`supernatant was determined by measuring the ab-
`sorbance at 276 nm as described. Average absor-
`bance of the supernatant from tubes containing blank
`microspheres treated identically was subtracted from
`all measurements.
`
`2 .5. Scanning electron microscopy
`
`Microsphere surface structure and porosity were
`investigated
`by
`scanning
`electron microscopy
`(Hitachi S-4700). Samples were prepared by placing
`a droplet of an aqueous microsphere suspension onto
`a silicon stub. The samples were dried overnight and
`were sputter coated with gold prior to imaging at
`2–10 eV.
`
`2 .6. Particle size distribution
`
`A Coulter Multisizer 3 (Beckman Coulter)
`equipped with a 100- or 280-mm aperture was used
`to determine the size distribution of the various
`sphere preparations. The lyophilized particles were
`resuspended in Isoton electrolyte and a type I-A
`
`Page 3 of 11
`
`

`

`140
`
`C. Berkland et al. / Journal of Controlled Release 82(2002)137–147
`
`dispersant was used to prevent microsphere aggrega-
`tion. A minimum of 5000 microspheres was ana-
`lyzed for each sample.
`
`3 . Results
`
`3 .1. Microsphere fabrication and characterization
`
`Uniform PLG microspheres were fabricated em-
`ploying the spraying apparatus described previously
`[20]. The model drug compounds, rhodamine B and
`piroxicam (free base form), were encapsulated by
`co-dissolving the drug with the PLG in DCM. In
`order to examine the effect of microsphere diameter
`on drug release kinetics, we fabricated rhodamine-
`containing particles of 20, 40 and 65 mm, at theoret-
`ical loadings of 1, 3, and 5%, and piroxicam-con-
`taining particles of 10, 50 and 100 mm with 5, 10,
`15, and 20% loading. Rhodamine and piroxicam
`loading and encapsulation efficiency (e.e.) are re-
`ported in Tables 1 and 2, respectively. (While drug
`loading is less than theoretical, e.e.510–60%, for
`simplicity we will refer to the various samples by the
`theoretical loading.)
`The microspheres were very uniform,
`
`typically
`
`Table 1
`Characterization of rhodamine-loaded PLG microspheres
`
`Fig. 1. Typical size distributions of uniform microspheres loaded
`with rhodamine (20-, 45- and 75-mm diameter) and piroxicam
`(10- and 55-mm diameter). All distributions are normalized by
`total area under the curve. Thus,
`the peak height
`is also an
`indication of the relative particle uniformity. Each distribution is
`colored with a different shade of gray to distinguish where they
`overlap.
`
`having .90% of the particles within 2-mm of the
`average diameter (Fig. 1). Microsphere homogeneity
`is also evident in scanning electron micrographs of
`the various microsphere preparations (Fig. 2). The
`
`20 mm
`
`1
`
`3
`
`5
`
`40 mm
`
`1
`
`3
`
`5
`
`65 mm
`
`1
`
`3
`
`5
`
`0.63
`
`1.80
`
`2.50
`
`0.37
`
`1.05
`
`1.75
`
`0.61
`
`1.29
`
`3.00
`
`63
`
`60
`
`50
`
`37
`
`35
`
`35
`
`61
`
`43
`
`60
`
`Theoretical
`loading (%)
`
`Experimental
`loading (%)
`
`Encapsulation
`efficiency (%)
`
`Table 2
`Characterization of piroxicam-loaded PLG microspheres
`
`10 mm
`
`f5
`
`10
`
`15
`
`20
`
`3.0
`
`4.6
`
`5.6
`
`5.8
`
`50 mm
`
`5
`
`1.0
`
`10
`
`15
`
`20
`
`1.0
`
`1.5
`
`3.6
`
`100 mm
`
`5
`
`1.0
`
`10
`
`15
`
`20
`
`3.1
`
`3.0
`
`5.8
`
`59
`
`46
`
`37
`
`29
`
`19
`
`10
`
`10
`
`18
`
`20
`
`31
`
`20
`
`29
`
`Theoretical
`loading (%)
`
`Experimental
`loading (%)
`
`Encapsulation
`efficiency (%)
`
`Page 4 of 11
`
`

`

`C. Berkland et al. / Journal of Controlled Release 82(2002)137–147
`
`141
`
`Fig. 2. Scanning electron micrographs of (A) 65- (B) 40- and (C) 20-mm rhodamine-loaded microspheres and (D) 100- (E) 50- and (F)
`10-mm piroxicam-loaded microspheres. Scale bar represents 100 mm.
`
`particles exhibit a smooth, slightly porous surface
`and dense polymer interior similar to microspheres
`produced using conventional emulsion techniques
`(Fig. 3) [16,17,23,24]. The average sizes determined
`by the Coulter counter are |5–10% larger than the
`sizes obtained from SEM, but
`the uniformity is
`readily apparent. The larger size may be the result of
`
`to the
`swelling due to water uptake. We refer
`particles according to the smaller sizes obtained from
`SEM.
`
`3 .2. In vitro release from uniform microspheres
`
`To examine the effect of microsphere size and size
`
`Page 5 of 11
`
`

`

`142
`
`C. Berkland et al. / Journal of Controlled Release 82(2002)137–147
`
`Fig. 3. Scanning electron micrographs of a fractured PLG micro-
`sphere depicting the dense polymer matrix of drug-encapsulating
`microspheres. Scale bars represents 10 mm.
`
`uniformity on drug release kinetics, we measured
`release profiles for both drugs from spheres of all
`loadings and sizes. Rhodamine release profiles are
`shown in Fig. 4. As expected, 20-mm microspheres
`exhibited a faster initial release than 65-mm micro-
`spheres, likely due to the increased surface-to-vol-
`ume ratio of the smaller particles. Further, as drug
`loading increased,
`the initial rate of drug release
`increased. An interesting concave-upward (i.e. sig-
`moidal) profile was observed with the 65-mm par-
`ticles and to a lesser extent with the 45-mm particles,
`wherein drug release was initially slow, then pro-
`gressed to a more rapid release phase before leveling
`off [25,26].
`trends
`Piroxicam release profiles show similar
`(Fig. 5). Samples of 10-, 50- and 100-mm micro-
`spheres were studied. The microspheres span a
`broader size range than the rhodamine-loaded par-
`ticles, resulting in a more pronounced difference in
`
`Fig. 4. Effect of microsphere size and drug loading on rhodamine
`release rates: (A) 1%, (B) 3% and (C) 5% theoretical loading.
`
`drug release profiles. The smallest microspheres (10-
`mm diameter) exhibited a rapid initial rate of release,
`with 40–60% of encapsulated piroxicam released
`within the first 24 h. Initial release rates decreased
`with increasing microsphere diameter for all drug
`loadings examined. Further, the initial release rate
`decreased with increasing drug loading. Interestingly,
`the 50- and 100-mm particles exhibited sigmoidal
`release profiles similar to rhodamine release from
`65-mm microspheres.
`
`Page 6 of 11
`
`

`

`C. Berkland et al. / Journal of Controlled Release 82(2002)137–147
`
`143
`
`and 3:1, w/w) of 20-mm/5% and 65-mm/5%
`rhodamine-containing microspheres. The release pro-
`files for the microsphere mixtures were intermediate
`between the two individual microsphere release
`profiles (Fig. 6A). Further,
`the mixture release
`profiles corresponded to a mass-weighted linear
`combination of the individual release profiles (see
`dotted lines in Fig. 6A).
`Given the agreement between predicted release
`profiles (i.e. linear combinations) and the experimen-
`tal data, various linear combinations of individual-
`microsphere release profiles, based on the nine
`preparations of varying microsphere
`size
`and
`rhodamine loading, were examined to identify a
`combination of uniform microspheres that might
`
`Fig. 5. Effect of microsphere size and drug loading on piroxicam
`release rates: (A) 5%, (B) 10% and (C) 15% theoretical loading.
`
`3 .3. In vitro release from mixtures of uniform
`microspheres
`
`Based on the different shapes of the uniform
`microsphere release profiles, and given the repro-
`ducibility of our methodology for uniform micro-
`sphere fabrication, we reasoned that
`it may be
`possible to modulate release kinetics in a desired
`fashion by mixing appropriate proportions of two or
`more uniform microsphere preparations [3,10]. To
`test this hypothesis, we mixed known ratios (1:3, 1:1
`
`Fig. 6. (A) Rhodamine release from 20-mm /5% microspheres,
`65-mm/5% microspheres and 3:1, 1:1 and 1:3 (w/w) mixtures.
`(B) Rhodamine release from 20-mm/5% microspheres, 65-mm/
`3% microspheres and 1:4, 1:9 and 1:24 (w/ w) mixtures. Filled
`symbols: experimental data points for individual microspheres.
`Open symbols: experimental data points for mixtures. Dotted line:
`weighted average of individual microsphere experimental release
`data (predicted release). Error bars, typical of those shown in Fig.
`4, were removed for clarity.
`
`Page 7 of 11
`
`

`

`144
`
`C. Berkland et al. / Journal of Controlled Release 82(2002)137–147
`
`result in zero-order release kinetics. Based on the
`predicted release profiles, we chose to mix 20-mm /
`5% and 65-mm/3% microspheres in various ratios
`and performed in vitro release studies for comparison
`to predicted release profiles (Fig. 6B). Again, the
`experimental release data coincided precisely with
`the predicted release profile, and the 1:4 (w/ w)
`mixture provided constant release for 8 days.
`To investigate the generality of this approach for
`modulating drug release kinetics, we performed a
`similar set of experiments to generate zero-order
`release of the clinically relevant NSAID, piroxicam.
`Multiple linear combinations of 10-, 50- and 100-mm
`piroxicam-containing microspheres at various drug
`loadings were examined computationally to identify
`a combination resulting in linear drug release. Two
`possible formulations were found. The first formula-
`
`Fig. 7. (A) Piroxicam release from 3:1, 1:1 and 1:3 (w /w)
`mixtures of 10-mm/15% microspheres and 50-mm /15% micro-
`spheres. (B) Piroxicam release from 1:6.1, 1:11.5 and 1:39 (w /w)
`mixtures of 10-mm/20% microspheres and 50-mm /10% micro-
`spheres.
`
`tion combined 10-mm/15% and 50-mm/15% micro-
`spheres in ratios of 3:1, 1:1, and 1:3 (w/w). This
`formulation resulted in slightly concave downward
`release profiles for the 3:1 and 1:1 ratios and a linear
`drug release profile for the 1:3 (w/w) mixture (Fig.
`7A). The second formulation comprised 10-mm/ 20%
`and 50-mm/10% microspheres in three ratios, 1:6.1,
`1:11.5, and 1:39 (w/w). The 1:6.1 (w/w) formulation
`produced near linear release over the course of the
`experiment (Fig. 7B).
`
`4 . Discussion
`
`Long-term zero-order release of small-molecule
`therapeutics from biodegradable microspheres has
`been difficult
`to achieve. Release of model com-
`pounds similar to rhodamine B is often rapid and
`diffusion controlled [27,28]. Similarly, release of
`NSAIDs, especially piroxicam, encapsulated in poly-
`meric particles typically occurs within 24 h and is
`dominated by a large initial
`rate of
`release (or
`‘burst’), offering little advantage over conventional
`oral dosage forms [16,17,23,24,29,30]. In contrast,
`our
`results show that simple molecules can be
`released in a controlled manner over significant
`durations of time.
`The initial release rates of both rhodamine and
`piroxicam decreased with increasing sphere diam-
`eter. This is expected due to the decrease in surface
`area/volume ratio with increasing size. Furthermore,
`we observed that the release rates (as percent of total
`drug released vs. time) from rhodamine-containing
`microspheres increased slightly with increasing load-
`ing for all sphere sizes. For purely diffusion-con-
`trolled release, no dependence on drug loading is
`expected. Other
`researchers have reported faster
`release with increased loading, especially for cases in
`which the drug is phase-separated from the polymer
`matrix and release can occur through aqueous meso-
`and macropores created by the drug [31,32]. Such a
`mechanism may explain the effect of rhodamine
`loading on release rates. In contrast, we observed
`that
`the release rate of piroxicam decreased with
`increasing loading. Possible explanations for this
`surprising result may be that the piroxicam, with pKa
`of 5.07 and 2.33, is buffering acidification of the
`intrapolymer environment caused by accumulation of
`
`Page 8 of 11
`
`

`

`C. Berkland et al. / Journal of Controlled Release 82(2002)137–147
`
`145
`
`the hydro-
`polymer degradation products, or that
`phobic drug is slowing water uptake. Another possi-
`bility is that, due to the low solubility of piroxicam,
`‘sink’ conditions were not achieved. The highest
`piroxicam concentrations measured were |100 mg /
`ml while the solubility of piroxicam in the release
`media at 37 8C is 1.3 mg/ ml (the relatively high
`solubility is due to the presence of Tween 20 at a
`concentration, 0.5%, above its critical micelle con-
`centration). Because the maximum piroxicam con-
`centration was less than one tenth of its solubility,
`drug concentration in the release media likely had
`little effect on the release kinetics.
`Small and large particles resulted in qualitatively
`different release profiles. The smallest microspheres
`(10- and 20-mm diameter, encapsulating piroxicam
`and rhodamine,
`respectively) exhibited concave
`downward release profiles typical of diffusion-con-
`trolled release from PLG microspheres. However, the
`larger microspheres exhibited a sigmoidal release
`profile in which the release rate initially increased
`with time. Similar release profiles have been ob-
`served previously. For example, Sansdrap and Moes
`reported a sigmoidal release profile for a low solu-
`bility (10 mg/ml
`in water) drug, nifedipine [21]
`using 80-mm (630 mm) PLG microspheres, while
`Guiziou et al. found sigmoidal release from poly(lac-
`tic acid) (PLA) microspheres loaded with the NSAID
`indomethacin [24]. Further, Ravivarapu et al. re-
`ported a similar release profile for leuprolide acetate-
`loaded PLG microspheres [33].
`The sigmoidal release profiles may result from
`several mechanisms. Guiziou et al. attributed the
`upward bending release profile to the microspheres
`having a more dense polymer matrix with increased
`stability as compared to microspheres fabricated
`using lower molecular weight PLA which released in
`a more diffusion-controlled manner [24]. Alterna-
`tively, water uptake, resulting in increased solubiliza-
`tion of the drug and swelling of the polymer matrix
`[21], may cause an increasing release rate. However,
`we propose that the sigmoidal shape results from
`polymer degradation. Erosion was suggested as the
`cause of sigmoidal
`leuprolide release profiles de-
`scribed above [33]. Polymer hydrolysis, with or
`without
`the subsequent mass loss,
`is expected to
`cause both increasing diffusivity of drug through the
`polymer matrix and an increase in the size of
`
`aqueous pores in the polymer (also potentially lead-
`ing to an increase in the effective diffusivity).
`Additional studies directly addressing this hypothesis
`are in progress.
`Constant release is highly desirable for many drug
`delivery applications. Because there is a transition
`from the concave downward to sigmoidal release
`profiles as sphere size increases,
`it appears that
`nearly linear release may be achieved at a certain
`size. For example, between 10 and 50 mm, a
`microsphere size may exist that would provide zero-
`order piroxicam release over a 4- to 8-day duration
`(cf. Figs. 4 and 5). Others report linear or near-linear
`release profiles achieved with microspheres of simi-
`lar size, |30–50 mm in diameter
`[4,5,10]. For
`example, Woo et al. formulated a leuprolide delivery
`system using PLA microspheres with an average
`diameter of 51.7 mm achieving near-linear peptide
`release for 135 days following a 15-day period of
`‘diffusion-controlled release’
`[5]. Our hypothesis
`suggests that the early phase of release results from
`the portion of the microspheres in this formulation
`under |35 mm, which would be expected to release
`drug more rapidly. Further, Bezemer et al. used a
`poly(ethylene glycol)-poly(butylene
`terephthalate)
`(PEG-PBT) block copolymer to test the effects of
`microsphere size on drug release [10]. They also
`discovered that decreasing the average microsphere
`size from 108 to 29 mm causes the release kinetics to
`change gradually from zero-order release to release
`controlled by Fickian diffusion. The microspheres
`used for these experiments were not uniform, but the
`trends are indicative of the trends we observed for
`uniform PLG microspheres.
`Other researchers have suggested that drug deliv-
`ery rates may be controlled by mixing microspheres
`of varying sizes or characteristics. For example,
`Ravivarapu et al. mixed microspheres comprising
`8.6- or 28.3-kDa PLG encapsulating leuprolide ace-
`tate [33]. The low-molecular-weight polymer
`re-
`sulted in porous, quickly releasing microspheres
`while the high-molecular-weight formulation resulted
`in dense microspheres and produced a sigmoidal
`release profile. By mixing microspheres comprising
`the two polymers, release rates could be tailored, and
`the resulting profiles were linear combinations of
`those
`resulting
`from individual microspheres.
`Bezemer et al. produced linear lysozyme release over
`
`Page 9 of 11
`
`

`

`146
`
`C. Berkland et al. / Journal of Controlled Release 82(2002)137–147
`
`25 days from PEG-PBT microspheres having a
`bimodal size distribution dominated by 50- and 110-
`mm particles (in essence a combination of two sizes)
`[10]. Finally, Narayani and Panduranga Rao com-
`bined gelatin microspheres of various size ranges
`producing zero-order release of methotrexate [3].
`Because release kinetics from uniform spheres are
`very predictable and reproducible, our ability to
`fabricate uniform microspheres enhances such a
`technique. We found that upon mixing uniform
`microsphere preparations, the resulting release pro-
`file is a mass-weighted average of the release profiles
`of the individual microspheres. This demonstrates
`that the microspheres release their payload indepen-
`dently; there is no interaction between the particles.
`In these experiments, the shapes of the rhodamine
`and piroxicam release profiles were such that it was
`possible to choose appropriate microsphere mixtures
`that provided zero-order release kinetics (Figs. 6 and
`7). However,
`it may not always be possible to
`generate a desired release profile from mixtures of
`only two microsphere sizes. Depending on the
`desired profile and the shape of the individual release
`curves, one may need to mix multiple microsphere
`samples or to fabricate complex microsphere size
`distributions. Because the reported fabrication meth-
`od provides a unique ability to generate predefined
`microsphere sizes [20], this technology may lead to
`enhanced control of release rates.
`
`5 . Conclusions
`
`Microsphere size is a primary determinant of drug
`release kinetics. Release of model small-molecule
`drugs can be varied from typical diffusion-controlled
`profiles to slower, sigmoidal profiles as microsphere
`diameter is increased in the range of 10–100 mm.
`Drug release from mixtures of uniform microspheres
`corresponds to a weighted average of the release
`from individual uniform microspheres. As a result, it
`is possible to choose appropriate mixtures to gener-
`ate desired release rate profiles, in particular constant
`release. Thus, microsphere mixtures with well-de-
`fined size distributions may provide a general meth-
`odology for controlling drug release rates.
`
`A cknowledgements
`
`We wish to thank Young Bin Choy for improve-
`ments to the microparticle fabrication apparatus and
`Dong Wha Pharmaceuticals (Korea) for providing
`piroxicam.
`
`R eferences
`
`[1] D.M. Schachter, J. Kohn, A synthetic polymer matrix for the
`delayed or pulsatile release of water-soluble peptides, J.
`Controlled Release 78 (2002) 143–153.
`[2] D.L. Wise, D.J. Trantolo, R.T. Marino, J.P. Kitchell, Oppor-
`tunities and challenges in the design of implantable bio-
`degradable polymeric systems for the delivery of anti-micro-
`bial agents and vaccines, Adv. Drug Deliv. Rev. 1 (1987)
`19–39.
`[3] R. Narayani, K. Panduranga Rao, Gelatin microsphere
`cocktails of different sizes for the controlled release of
`anticancer drugs, Int. J. Pharm. 143 (1996) 255–258.
`[4] R.T. Liggins, S. D’Amours, J.S. Demetrick, L.S. Machan,
`H.M. Burt, Paclitaxel
`loaded poly(L-lactic acid) micro-
`spheres for the prevention of intraperitoneal carcinomatosis
`after a surgical repair and tumor cell spill, Biomaterials 21
`(2000) 1959–1969.
`[5] B.H. Woo, J.W. Kostanski, S. Gebrekidan, B.A. Dani, B.C.
`Tahanoo, P.P. DeLuca, Preparation, characterization and in
`vivo evaluation of 120-day poly(DL-lactide) leuprolide micro-
`spheres, J. Controlled Release 75 (2001) 307–315.
`[6] A.M. Radder, H. Leenders, C.A. van Blitterswijk, Applica-
`tion of PEO/PBT copolymers for bone replacement, J.
`Biomed. Mater. Res. 30 (1996) 341–351.
`[7] B. Lemmer, in: R. Gurny, H. Junginger, N. Peppas (Eds.),
`Why Are So Many Biological Systems Periodic, Wis-
`senschaftliche Verlagsgesellschaft, Stuttgart, 1992, pp. 11–
`24.
`[8] J.W. Shiver, T.-M. Fu, L. Chen et al., Replication-incompe-
`tent adenoviral vaccine vector elicits effective anti-immuno-
`deficiency-virus immunity, Nature 415 (2002) 331–335.
`[9] J.L. Cleland, Single-administration vaccines: controlled-re-
`lease technology to mimic repeated immunizations, TIBtech
`17 (1999) 25–29.
`[10] J.M. Bezemer, R. Radersma, D.W. Grijpma, P.J. Dijkstra,
`C.A. van Blitterswijk, J. Feijen, Microspheres for protein
`delivery prepared from amphiphilic multiblock copolymers
`2. Modulation of release rate, J. Controlled Release 67
`(2000) 249–260.
`[11] Y. Men, C. Thomasin, H.P. Merkle, B. Gander, G. Corradin,
`A single administration of tetanus toxoid in biodegradable
`microspheres elicits T cell and antibody responses similar or
`superior to those obtained with aluminum hydroxide, Vaccine
`13 (1995) 683–689.
`[12] S. Cohen, T. Yoshika, M. Lucarelli, L.H. Hwang, R. Langer,
`
`Page 10 of 11
`
`

`

`C. Berkland et al. / Journal of Controlled Release 82(2002)137–147
`
`147
`
`Controlled delivery systems for proteins based on poly(lac-
`tic/glycolic acid) microspheres, Pharm. Res. 8 (1991) 713–
`720.
`[13] J. Akbuga, Effect of microsphere size and formulation
`factors on drug release from controlled-release furosemide
`microspheres, Drug Dev. Ind. Pharm. 17 (1991) 593–607.
`[14] Y. Ozalp, N. Ozdemir, S. Kocagoz, V. Hasirci, Controlled
`release of vancomycin from biodegradable microcapsules, J.
`Microencapsul. 18 (2001) 89–110.
`[15] S. Cohen, T. Yoshioka, M. Lucarelli, L.H. Hwang, R.
`Langer, Controlled drug delivery systems for proteins based
`on poly(lactic/glycolic acid) microspheres, Pharm. Res. 8
`(1991) 713–720.
`[16] M. Tuncay, S. Calis, H.S. Kas, M.T. Ercan, I. Peksoy, A.A.
`Hincal, Diclofenac sodium incorporated PLGA (50:50)
`microsphe