Tpo

HMa

b

c

d

e

a

ARRAA

KMBSPPI

1

cHedthtdd

es

0d

Colloids and Surfaces B: Biointerfaces 74 (2009) 340–349

Contents lists available at ScienceDirect

Colloids and Surfaces B: Biointerfaces

journa l homepage: www.e lsev ier .com/ locate /co lsur fb

he effect of formulation variables on the characteristics of insulin-loadedoly(lactic-co-glycolic acid) microspheres prepared by a single phaseil in oil solvent evaporation method�

amed Hamishehkara, Jaber Emamib, Abdolhossein Rouholamini Najafabadic, Kambiz Gilani c,ohsen Minaiyanb, Hamid Mahdavid, Ali Nokhodchie,∗

Pharmaceutical Technology Laboratory, Drug Applied Research Center, Tabriz University (Medical Sciences), Tabriz, IranSchool of Pharmacy and Pharmaceutical Sciences and Isfahan Pharmaceutical Research Center, Isfahan University of Medical Sciences, Isfahan, IranAerosol Research Laboratory, School of Pharmacy, Tehran University of Medical Sciences, Tehran, IranDepartment of Novel Drug Delivery Systems, Iran Polymer and Petrochemical Institute, Tehran, IranMedway School of Pharmacy, The Universities of Kent and Greenwich, Central Ave., Chatham Maritime ME4 4TB, Kent, UK

r t i c l e i n f o

rticle history:eceived 15 April 2009eceived in revised form 18 May 2009ccepted 4 August 2009vailable online 12 August 2009

eywords:icroencapsulation

iodegradable microspheres

a b s t r a c t

Biodegradable polymeric microspheres are ideal vehicles for controlled delivery applications of drugs,peptides and proteins. Amongst them, poly(lactic-co-glycolic acid) (PLGA) has generated enormous inter-est due to their favorable properties and also has been approved by FDA for drug delivery. Insulin-loadedPLGA microparticles were prepared by our developed single phase oil in oil (o/o) emulsion solvent evap-oration technique. Insulin, a model protein, was successfully loaded into microparticles by changingexperimental variables such as polymer molecular weight, polymer concentration, surfactant concentra-tion and stirring speed in order to optimize process variables on drug encapsulation efficiency, releaserates, size and size distribution. A 24 full factorial design was employed to evaluate systematically the

olvent evaporationLGArotein deliverynsulin

combined effect of variables on responses. Scanning electron microscope (SEM) confirmed sphericalshapes, smooth surface morphology and microsphere structure without aggregation. FTIR and DSC resultsshowed drug–polymer interaction. The encapsulation efficiency of insulin was mainly influenced by sur-factant concentration. Moreover, polymer concentration and polymer molecular weight affected burstrelease of drug and size characteristics of microspheres, respectively. It was concluded that using PLGA

eightof ins

with higher molecular wencapsulation efficiency

. Introduction

Significant advances in biotechnology have resulted in the dis-overy of a large number of therapeutic and antigenic proteins [1].owever, their unique properties, such as high molecular weight,asy degradation, instability and low bioavailability, make tra-itional dosage forms not proper to deliver them [2]. Insulin ishe most important regulatory hormone in the control of glucose

omeostasis [3]. Insulin replacement is essential in all patients withype I diabetes and in an increasing number of patients with type 2iabetes. However, the initiation of insulin treatment is frequentlyelayed in the latter population and, in general, associated with

� This work was presented, in part, at the 2007 International Pharmaceutical Fed-ration (FIP) in Beijing, China and awarded as the best poster in Industrial Pharmacyection.∗ Corresponding author. Tel.: +44 1634 883846; fax: +44 1634 883927.

E-mail address: a.nokhodchi@kent.ac.uk (Ali Nokhodchi).

927-7765/$ – see front matter © 2009 Elsevier B.V. All rights reserved.oi:10.1016/j.colsurfb.2009.08.003

, high surfactant and polymer concentrations led to a more appropriateulin with low burst effect and desirable release pattern.

© 2009 Elsevier B.V. All rights reserved.

poor compliance in both populations, as subcutaneous adminis-tration is inconvenient and unacceptable by many patients andhealth care providers. Usually, insulin is injected subcutaneouslytwo to four times a day [4]. Therefore, there has been significantinterest in the development of an insulin formulation that couldrelease the drug in a controlled fashion for longer periods. Themost promising approach to sustain the release of peptide andproteins is their encapsulation within microparticles composed ofbiodegradable and biocompatible polymers [5]. Both natural andsynthetic biodegradable polymers have been investigated for con-trolled drug release [6,7]. In this sense, there is a particular interestin poly(lactide-co-glycolide) (PLGA) (the copolymer of lactic andglycolic acids) microspheres and a wealth of literature has beengenerated on this issue [8]. PLGAs have shown to be biocompatible

and they degrade to toxicologically acceptable lactic and glycolicacids [9]. Also, they have been approved by FDA as controlleddrug release microspheres [10]. However, despite this great poten-tial, there are still considerable difficulties that have limited theuse of this system in the commercial market. Low encapsulation

urfaces B: Biointerfaces 74 (2009) 340–349 341

easp(tteoweoapIhaamfclssm(ralw

2

2

1m5idCawlwu

2

lphp3miIeg3rst

Table 1Factors and factor levels investigated in this study according to a full factorial exper-imental design.

Factors Levels

A Type of PLGAa 502Hb 504Hc

B PLGA concentration (%) 1.6 3.2C Surfactant concentration (%) 0.3 3.0D Stirring speed (rpm) 3000 5000

a

H. Hamishehkar et al. / Colloids and S

fficiency, protein inactivation during the encapsulation processnd difficulties for controlling the release of the active protein areome difficulties associated with encapsulation of peptides androteins using routine methods [11] including phase-separationcoacervation) [12], spray-drying [13] and solvent evaporationechniques (w/o, w/o/w, s/o/w and s/o/o) [14]. We investigatedhat devising and developing a single phase oil in oil (o/o) solventvaporation technique for preparation of PLGA microspheres couldvercome the aforementioned limitations. The purpose of the studyas, therefore, to design and develop a single phase o/o solvent

vaporation method applicable to the pharmaceutical productionf insulin-loaded PLGA microparticles satisfying pharmaceuticallycceptable criteria with respect to microsphere size distribution,rotein loading, adjustable release profiles and low burst release.

n this method, insulin and PLGA was simultaneously dissolved in aomogenous single phase composed of an appropriate mixture ofcetonitrile and water. Therefore, in the present article, formulationnd process variables affecting the characteristics of insulin-loadedicrospheres fabricated by developed method are reported. A 24

ull factorial design was used to systematically investigate theombined influence of four variables including polymer molecu-ar weight, polymer concentration, surfactant concentration andtirring speed on the drug encapsulation efficiency, release rates,ize and size distribution of the microspheres. Insulin-loaded PLGAicrospheres were characterized by Fourier transform infrared

FTIR) spectroscopy, differential scanning calorimetry (DSC), X-ay diffraction (XRD), and scanning electron microscope (SEM). Inddition, in the course of preparation and release studies of insulin-oaded PLGA microparticles, the stability of encapsulated protein

as fully evaluated and published elsewhere [15].

. Materials and methods

.1. Materials

PLGA polymers (RG502H, lot # 1006825 and RG504H, lot #020751) were supplied by Boehringer Ingelheim, Germany. Theole ratios of glycolic acid to l-lactic acid in the polymers were

0:50. These polymers have an uncapped free carboxylic acid groupn the terminal end. Recombinant human insulin (27.5 U/mg onried bases by HPLC) was kindly supplied by Exir Pharmaceuticalompany. Iran. Viscous paraffin oil and Span 80 (Merck, Germany)nd acetonitrile and hexane (both HPLC grade, Merck, Germany)ere used as received. Trifluoroacetic acid (TFA) (Fluka, Switzer-

and) was used in HPLC analysis. All other chemicals and reagentsere of analytical grade, purchased from commercial vendors andsed as received.

.2. Microsphere preparation

Insulin-loaded microspheres were prepared at the theoreticaloading of 6.25 and 12.50% of insulin (6.25 and 12.50 mg of insuliner 100 mg of polymer). Insulin crystals were dissolved in 0.01Nydrochloric acid (HCl) to give a concentration of 10 mg/ml. Therotein solution was added to PLGA solution in acetonitrile (1.6 or.2%) at V/V ratio of 1:5. This solution was then dispersed into 30 mlineral oil in the presence of Span 80 (0.3 or 3%) and stirred with an

mpeller type stirrer (Tohid Sanat Sepahan, Model TSS55, Isfahan,ran) at two different stirring speeds (3000 or 5000 rpm) for 2 h to

vaporate of inner phase. Microspheres were collected by centrifu-ation (Sigma 3K30, Rotor No. 12150, Germany) at 20,000 rpm for0 min at 10 ◦C and washed four times with n-hexane to completeemove of mineral oil. Particles were filtered, vacuum dried, andtored under refrigeration in a desiccator until used. The prepara-ion conditions and compositions are summarized in Table 1.

Poly(lactic-co-glycolic acid).b RG502H.c RG504H.

For more information refer to Section 2.1.

2.3. Experimental design

Traditionally, pharmaceutical formulations have been devel-oped by changing one variable at a time. The method is timeconsuming and requires lots of imaginative efforts. Moreover, itmay be difficult to evolve an ideal formulation using this classi-cal technique, since the combined effects of independent variablesare not considered. It is therefore, essential to understand the com-plexity of the pharmaceutical formulations by using the establishedstatistical tools such as factorial designs [16]. The factorial designexperiments were performed in random order and then variousformulations of the microspheres were developed. The designedmatrix from a single replicate of 24 experiments along with theformulation codes are given in Table 2. The variables (factors)studied were as follows: polymer type (502H and 504H) whichdiffer in molecular weight, polymer concentration in acetonitrile(1.6 or 3.2%), stirrer speed (3000 or 5000 rpm) and surfactant(Span 80) concentration in oil phase (liquid paraffin) (0.3 or 3%).The factor levels were chosen in accordance with the results ofour preliminary studies a portion of which is recently reported[17]. The relevant and important features of microspherical sys-tems such as encapsulation efficiency, burst release, size and sizedistribution were evaluated as the responses. The aim was toachieve low burst release, high encapsulation efficiency of insulin,narrow size distribution of prepared microspheres and desiredsustained release pattern of insulin. To perform the statistical anal-yses of the data, the Minitab® 14 software (Minitab Inc., USA) wasemployed.

2.4. Polymer molecular weight analysis

The molecular weight of the polymers was determined by gelpermeation chromatography (GPC) using an Agilent 1100 Serieswith refractive index detector (Agilent Technologies, USA). Col-umn temperature was maintained at 23 ◦C. Samples were filteredand eluted in tetrahydrofuran through a series of 102, 103 and 104PLgel Agilent columns at a flow rate of 1 ml/min. Weight and num-ber average molecular weights were calculated from a calibrationcurve constructed on the basis of a series of polystyrene standardswith molecular weight ranging from 1350 to 151700 (Polymer Lab-oratories Inc., MA, USA).

2.5. Insulin encapsulation efficiency measurement

Accurately weighed 5 mg of microspheres were first dissolved,under mild agitation (vortex), in 300 �l of acetonitrile in testtubes. Then, 1.7 ml of 0.01N HCl was added dropwise to theabove vortexed mixture to dissolve insulin completely. The mixture

was allowed to settle for 30 min while the polymer precipi-tated. The remaining undissolved PLGA was then separated bycentrifugation (10,000 rpm for 2 min, Sigma 1-13, Germany). Theclear supernatants were then withdrawn and analyzed for insulincontent using a reversed-phase HPLC Waters 2690 separations

342 H. Hamishehkar et al. / Colloids and Surfaces B: Biointerfaces 74 (2009) 340–349

Table 2Run Parameters for two-level four-factorial experimental design.

Experiment Polymer type Polymer concentration (%) Stirring speed (rpm) Surfactant concentration (%)

1 502H 3.2 5000 3.02 502H 3.2 5000 0.33 502H 1.6 3000 0.34 502H 3.2 3000 3.05 504H 3.2 5000 3.06 504H 1.6 5000 3.07 504H 3.2 3000 3.08 502H 3.2 3000 0.39 502H 1.6 5000 3.0

10 504H 1.6 3000 3.011 504H 3.2 3000 0.312 504H 3.2 5000 0.313 502H 1.6 3000 3.014 504H 1.6 5000 0.3

ma

aWsT1atef

E

2

uaatasutfcw

C

wtl

2

ataSowT

2.10. X-ray diffraction (XRD)

15 502H 1.616 504H 1.6

odule equipped with a UV–visible detector Waters 2487 dual �bsorbance detector (Waters, Milford, MA, USA).

Chromatographic separation was performed at room temper-ture using a C18 �-Bondapak (250 mm × 4.6 mm, 10 �m, 125 Åaters, Ireland) chromatographic column. The mobile phase con-

isted of acetonitrile: water mixture (35:65) containing 0.1% TFA.he flow rate and sample injection volume were 1.5 ml/min and00 �l, respectively. The absorbance of insulin was detected atwavelength of 214 nm. Insulin solutions of known concentra-

ions (1–80 �g/ml) were used to generate calibration curves. Thencapsulation efficiency of the microspheres was calculated by theollowing equation:

ncapsulation efficiency (%, w/w) = actual drug contentnominal drug content

× 100

.6. Insulin in vitro release study

The PLGA microspheres, 5 mg of each formulation, were individ-ally suspended in 2 ml phosphate buffer 20 mM pH 7.4 in test tubesnd incubated in a shaker (Grant instruments, Cambridge, England)t 37 ◦C and 50 rpm. At the time intervals of 1, 6, 12, 24, 48, and 72 h,he sample microtubes were centrifuged, 40 �l of supernatant wasccurately withdrawn, and replaced with 40 �l of fresh buffer. Theamples were then analyzed using the reverse-phase HPLC systemnder the same operating conditions as described in the encapsula-ion efficiency experiment. The microspheres were resuspended inresh buffer after sampling. The in vitro release experiments wereonducted in triplicate. The cumulative amount of insulin releasedas calculated using the following equation:

umulative amount of insulin released (%) = Mt

M∞× 100

here Mt is amount of insulin released at time t and M∞ is theotal amount of insulin released at time infinity, which is the actualoading of insulin determined in loading efficiency experiment.

.7. Particle size and morphology

The surface morphology and size of the microspheres weressessed by a scanning electron microscope (SEM), Philips XL30,he Netherlands. The powder sample was spread on a SEM stub

nd sputtered with gold using sputter coater SCDOOS, BAL-TEC,witzerland. Particle size was obtained by measuring the diametersf at least 300 particles shown in SEM using image analysis soft-are (Image-Pro Plus 4.5; Media Cybernetics, Silver Spring, USA).

he size of each sample was measured at triplicate and the mean

5000 0.33000 0.3

value was calculated. The size distribution was evaluated with thespan value defined as follows:

Span = D90% − D10%

D50%

where DN% (N = 10, 50, 90) is the volume percentage of microsphereswith diameters up to DN% is equal to N%. The smaller span valueindicates the narrower size distribution.

2.8. Fourier transform infrared (FTIR)

Insulin, PLGA and insulin-loaded microspheres were mixed withKBr and were pressed to disk. Infrared (IR) spectra of the sampleswere scanned in the range from 400 to 4000 cm−1 and recordedon a Fourier transform infrared spectrometer (Equniox 55 LS 101Bruker, German) to study the possible interactions between insulinand PLGA. FTIR spectra were obtained at a resolution of 4 cm−1

with a minimum of 256 scan per spectrum. All measurements weretaken at room temperature. The spectra of water, CO2 and KBr weresubtracted from the sample spectrum and the procedure was doneunder nitrogen gas to prevent humidity interference.

2.9. Differential scanning calorimetry (DSC)

Glass transition temperatures (Tg) of the polymers, insulin andthe insulin-loaded microspheres were measured with a differentialscanning calorimeter (DSC822e, Mettler Toledo, Switzerland). Sam-ples were prepared by carefully weighing 7–8 mg of microspheresinto an aluminum pan and then hermetically sealed. The DSCexperiment for unprocessed insulin was carried out with the equiv-alent amount of insulin loaded in the microspheres (0.3–0.4 mg).The pans were then heated at a rate of 10 ◦C/min from 0 to280 ◦C under a constant flow of nitrogen gas. Calibration of thesystem was performed using indium and zinc standards. The cal-culations were performed using the Mettler STARe version 8.01software.

The crystallinity of insulin, blank microspheres and extractedinsulin from microspheres were evaluated using X-ray diffrac-tometer (x-Pert, Philips, UK) with a Cu tube anode. The X-raydiffractogram was scanned with the diffraction angle increasingfrom 5◦ to 70◦, 2� angle.

H. Hamishehkar et al. / Colloids and Surfaces B: Biointerfaces 74 (2009) 340–349 343

icroparticles showing the external surface structure of PLGA microspheres.

3

3

wlmtsIsscppopmspTpav

3

m

Table 3Polymer molecular weight determined by GPCa.

Polymer Mwb Mnc PDId

502H 6065 3742.0 1.65504H 35813.5 23484.5 1.53

a Gel permeation chromatography.b

TO

Fig. 1. Scanning electron microscopy images of insulin-loaded PLGA m

. Results and discussion

.1. Preparation of insulin-loaded PLGA microspheres

Various formulations of insulin-loaded PLGA particulate systemere fabricated according to a full factorial experimental design

isted in Table 2. Microencapsulation by the solvent evaporationethod is, in principal, quite simple and involves two major steps,

he formation of stable droplets of the drug-containing polymerolution and the subsequent removal of solvent from the droplets.n practice, however, the reproducible manufacturing of micro-pheres with the desired properties (high encapsulation efficiency,uitable release profile, desired size and narrow size distribution),an be difficult, due to the large number of factors affecting theroperties of microspheres, such as polymer concentration, type ofolymer and its molecular weight, stirring speed, etc. The effectf each of these parameters has to be determined empirically,redictions and scale up remain a problem. Therefore, more infor-ation is needed in order to identify the relevant parameters and

ave development resources [14]. Discrete and relatively monodis-ersed insulin-loaded PLGA microspheres were prepared (Fig. 1a).he microspheres were spherical with smooth surface devoid ofores or cavities (Fig. 1b). The same appearance was observed forll formulations, independent of the microencapsulation processariables, as summarized in Table 1.

.2. Determination of the average molecular weight of polymer

Polymers 502H and 504H were characterized by GPC to deter-ine their weight (Mw) – and number (Mn) – average molecular

able 4btained experimental responses for different formulations (each number represents me

Experiment Encapsulation efficiency (%) Burs

1 101.16 ± 9.23a 68.02 20.82 ± 4.61 51.93 23.76 ± 0.23 75.94 86.76 ± 1.75 69.35 51.18 ± 6.85 27.86 49.53 ± 4.89 77.17 68.45 ± 7.55 24.48 28.75 ± 0.41 54.29 105.71 ± 0.87a 74.4

10 73.83 ± 6.72 64.011 13.72 ± 1.25 68.212 6.41 ± 0.46 58.313 26.20 ± 0.17 85.514 6.90 ± 0.69 74.215 12.98 ± 0.36 69.116 7.59 ± 1.77 75.0

a Insignificant overestimation might be due to experimental error particularly during i

Average molecular weights.c Average molecular number.d Polydispersity index.

Each sample was measured in duplicate.

weights and polydispersity index (PDI) (Table 3). All GPC chro-matographs contained one peak corresponding to the polymer. Nosmaller Mw fragments were evident. The low value of PDI indicatedthat the polymer chains possessed a low degree of polydispersity[18] and consequently the results on the effect of polymer molec-ular weight can be discussed with more trust.

3.3. Encapsulation efficiency

The percentage of encapsulation efficiency of insulin in allparticulate formulations were measured by HPLC and listed inTable 4. An increase in the encapsulation efficiency was observedin the microspheres prepared with PLGA with the lower molecu-lar weight, showing that PLGA with the lower molecular weight

effectively trapped insulin (Fig. 2a). This seems to be caused by theinteraction of carboxyl groups at the end of PLGA chains with thepositively charged amino groups of insulin. In this sense, it could beaccepted that although the exact concentration of the final carboxylgroups in these copolymers is not available, these groups would

an ± standard deviation, n = 3).

t release (%) Size (�m) Span

3 ± 1.29 1.54 ± 0.25 0.97 ± 0.109 ± 17.77 2.56 ± 0.54 1.27 ± 0.167 ± 8.32 1.69 ± 0.21 0.69 ± 0.051 ± 4.85 2.40 ± 0.23 0.53 ± 0.030 ± 3.96 2.45 ± 0.40 0.96 ± 0.090 ± 1.74 2.35 ± 0.38 0.95 ± 0.094 ± 5.07 3.92 ± 0.56 0.83 ± 0.072 ± 5.82 1.60 ± 0.19 0.68 ± 0.050 ± 7.07 1.04 ± 0.11 0.60 ± 0.047 ± 5.22 2.65 ± 0.45 1.01 ± 0.109 ± 0.39 8.47 ± 1.39 0.97 ± 0.090 ± 3.08 4.00 ± 0.70 1.02 ± 0.100 ± 10.68 1.53 ± 0.17 0.64 ± 0.052 ± 10.83 3.43 ± 0.92 1.63 ± 0.265 ± 8.04 2.31 ± 0.25 0.60 ± 0.041 ± 5.43 2.73 ± 0.48 1.04 ± 0.11

nsulin content determination by HPLC analysis.

344 H. Hamishehkar et al. / Colloids and Surfaces B: Biointerfaces 74 (2009) 340–349

F .6, 3.2E

nMp[dmiriethtt

Fr

ig. 2. Effect of variables including polymer type (502H, 504H) and concentration (1E (a); burst release (b); size (c); size distribution, span (d).

ormally be more numerous in the low Mw PLGA than in the highw PLGA [19]. Similar results and conclusions were reported for

ositively charged rifampicin [20], lysozyme [21] and asparaginase22]. This observation may indicate that, in the context of proteinelivery systems, special attention should be given to the poly-er composition and its affinity towards the protein. As illustrated

n Fig. 3a, surfactant concentration in the external oil phase hademarkable effect on encapsulation efficiency (P < 0.0001). Increas-ng surfactant concentration led to an increase in encapsulationfficiency of insulin in PLGA microspheres (Fig. 2a). Higher surfac-

ant concentration may result in a more stable emulsion whichinders the mass transfer of insulin with surroundings leadingo a more even and homogeneous distribution of protein withinhe interior of the PLGA microspheres. Such phenomenon is also

ig. 3. Paretro chart of standardized effects of polymer type, polymer concentration, stirelease, (b); size (c) and size distribution (d), ˛ = 0.05.

%), stirring speed (3000, 5000 rpm) and surfactant on the encapsulation efficiency,

suggested by Yang et al. working on bovine serum albumin PLGAmicrospheres [23]. Fig. 2a shows that a high polymer concentrationresulted in a higher encapsulation efficiency of insulin. This may beexplained by the fact that increase in drug–polymer phase viscos-ity, typically caused by higher concentration of the polymer, couldrestrict the migration of the drug to the continuous phase and thusimprove its entrapment [24]. These observations are in agreementwith previous reports [25,26]. Interactions between stirring speedand type of polymer and polymer concentration shown in Fig. 4amay be interpreted by this phenomenon. The effectiveness of poly-

mer concentration on encapsulation efficiency tended to be lessimportant if stirring speed increases. It was shown that an increasein stirring speed from 3000 to 5000 rpm itself had negligible effecton encapsulation efficiency (Figs. 2a and 3a).

ring speed and surfactant concentration on the encapsulation efficiency, (a); burst

H. Hamishehkar et al. / Colloids and Surfaces B: Biointerfaces 74 (2009) 340–349 345

releas

3

ifrpstwiilt

Fs

Fig. 4. Interaction plots for encapsulation efficiency (a); burst

.4. Insulin in vitro release study

The release profiles of insulin from all formulations are reportedn Fig. 5a and b. The release profiles exhibited an initial burst release,ollowed by a sustain insulin release pattern. The observed in vitroelease profile for insulin is a typical finding and has been reportedreviously with many alternative proteins loaded in PLGA micro-pheres [27–30]. The incomplete release may be due to the facthat microparticles had not yet entered the degradation phase inhich polymeric microparticles degrades and releases its remain-

ng drug content, or to incomplete detachment of the proteinnsulin adsorbed on the particle surface [5]. Incomplete release ofysozyme from PLGA microspheres has also been attributed to pro-ein adsorption to microparticle surfaces [31]. Burst release widely

ig. 5. Cumulative drug released from insulin-loaded PLGA biodegradable micro-pheres (formulations 1–16).

e (b); size (c); size distribution (d). For details refer to the text.

varied between 24 and 85% (Table 4 and Fig. 5a and b) which indi-cated that formulation variables played a major role in the releaseof insulin. Both polymer molecular weight and polymer concentra-tion in internal phase had significant effect on initial burst releaseof insulin from PLGA microspheres with the latter being moreeffective (Fig. 3b). Fig. 2b shows that increase in polymer molec-ular weight led to a decrease in burst release. These observationsare in accordance with the report that an increase in molecularweight of PLGA decreased tetanus toxoid release rate [32]. Diffu-sion rates of drug molecules and degradation products throughthe polymer phase generally decrease with increase in polymermolecular weight [33]. In addition, a major mechanism for releaseof many drugs is diffusion through water-filled pores formed byerosion. The pores are formed as polymer degrades and generatesmonomers and oligomers which are small enough to be solublein the surrounding medium and to diffuse out of the particles.These small products are formed more quickly following degra-dation of lower molecular weight polymers; in other words, fewerbonds must be hydrolyzed to generate soluble oligomers as ini-tial molecular weight decreases [21,26]. This will lead to higherrelease rate of protein. Cui et al. showed that there was a corre-lation between insulin release and the molecular weight of PLGA.An increase in the molecular weight of PLGA led to a reduction inboth the initial and final release over 24 h. This could be explainedby the retardant effect of the polymer meshwork generated bythe longer chains in the PLGA with higher molecular weight [34].Paretro chart shown in Fig. 3b indicated that polymer concentra-tion had the most significant impact on burst release (P = 0.008).Fig. 2b also shows that increasing in polymer concentration led todecreasing in burst release. In agreement with our results, polymerconcentration showed to be a key factor to influence the charac-teristics and release profiles of microspheres [25]. It was reportedthat the solidification of microspheres is more rapid at a high PLGAconcentration, which may result in a viscous polymer layer at themicrosphere droplet. It inhibits the diffusion of protein towards theexternal phase [26]. Stirring speed did not influence burst releaseconsiderably (Figs. 2b and 3b). Fig. 2b illustrates that increasing sur-

factant concentration of the external oil phase led to a decrease inthe burst release. This effect may be attributed to the high insulinencapsulation efficiency inside the PLGA microspheres as shownin Fig. 2a. Formulations 5 and 7 showed a low initial burst effectfollowed by 80% release with a near-zero order release kinetics

3 urface

aatlTweFttpt

46 H. Hamishehkar et al. / Colloids and S

nd a release rate constant of 1.05 (%/h) (Fig. 5a) which is desir-ble features satisfying the objective of the study. The polymerype and polymer and surfactant concentrations in these formu-ations were 504H, 3.2% and 3.0%, respectively (Tables 2 and 3).herefore, formulations containing PLGA with higher moleculareight and concentration and greater amount of surfactant (high-

st levels in the experimental design) showed lowest burst release.

urthermore, these two formulations released their protein con-ent almost entirely. These two formulations contained the sameype and amount of polymer and surfactant but they were pre-ared under two different stirring speeds. However, it was provenhat the stirring speed did not have any significant impact on the

Fig. 6. FTIR spectra of insulin (unprocessed), blank 502H and 504H PLGA micros

s B: Biointerfaces 74 (2009) 340–349

burst release (Figs. 2b and 3b). By comparing the slopes of thetwo graphed lines in a given graph the presence of any interactionbetween the two graphed variables can be visualized. The differ-ence in slope between the two lines represents the degree to whichthe two variables interact. That is, the effect of one factor is depen-dent upon a second factor. Fig. 4b shows that there is an interactionbetween surfactant concentration and polymer type. Because of dif-

ferences in the number of free carboxylic groups between 504H and502H PLGA, the hydrophilic–lipophilic balance of polymers wouldalso be different. Therefore, it can be suggested that the perfor-mance of the polymers would be affected by presence of surfactantand its concentration.

pheres and insulin-loaded PLGA microspheres (formulations No. 1 and 5).

urface

3

obbtaeipatTtpe

H. Hamishehkar et al. / Colloids and S

.5. Size characteristics of insulin-loaded PLGA microspheres

Particle size and size distribution are two important criteriaf microspheres as these factors affect [35] drug release rate,iodistribution, mucoadhesion, cellular uptake [36], water anduffer exchange to the interior of the microspheres, and pro-ein diffusion. The microspheres with a narrow size distributionre necessary to optimize the clinical outcomes [37]. Therefore,valuation of the effect of process variables on size character-stics of microspheres in order to develop a method that canrovide the uniform-sized microspheres composed of biodegrad-ble polymers was very imperative to us. The mean diameter of

he microspheres ranged from 1.04 to 8.47 �m as presented inable 4. It can be observed that there was a reduction in par-icle size when the surfactant concentration in the external oilhase was increased from 0.3 to 3.0% (V/V) (Fig. 2c). The pres-nce of surfactant in the external oil phase stabilizes emulsion

Fig. 7. DSC thermograms of insulin (unprocessed), blank 502H and 504H PLGA micr

s B: Biointerfaces 74 (2009) 340–349 347

droplets against coalescence, resulting in smaller emulsion droplets[23,36,38]. The size of the microspheres thus obtained is dependentupon the size and stability of the emulsion droplets formed dur-ing the stirring process [39]. Our results showed that an increasein the mixing speed led to a decrease in mean particle sizes(Fig. 2c). This might be due to the production of smaller emulsiondroplets through stronger shear forces and increased turbulenceas frequently reported about drug-loaded PLGA microparticles[23,40,41]. Polymer molecular weight and its concentration playeda major role on the size of PLGA microspheres (Fig. 3c). Poly-mer molecular weight had the most significant effect on the sizeand size distribution (P = 0.013 and P = 0.021, respectively). It was

reported that increasing in viscosity of the drug/matrix dispersiondue to increasing in PLGA molecular weight and concentrationyields larger microspheres because higher shear forces are nec-essary for droplet disruption [42–44]. The size distribution (spanvalues) of the samples ranged from 0.53 to 1.63 (Table 4). As

ospheres, insulin-loaded PLGA microspheres (formulations No. 1, 2, 5 and 6).

3 urfaces B: Biointerfaces 74 (2009) 340–349

sbsamcltt

3

coTpsif(amaboPbwrittu

3

tpeottevafiDmGmTrsiirmTTwhPh

The authors also would like to express their gratitude to Dr.Jamshidi for his technical assistance in GPC analysis and Dr. Imani

48 H. Hamishehkar et al. / Colloids and S

hown in Fig. 2d, size distributions of microspheres tended toe narrower with the low molecular weight polymer and highurfactant concentration. Fig. 4c indicates that there was an inter-ction between polymer concentration and stirring speed. Thiseans that the effectiveness of stirring speed and polymer con-

entration on the microspheres size are interrelated. It is clear thatower polymer concentration caused lower viscosities which inurn lead to more efficient effect of stirring speed on size reduc-ion.

.6. FTIR spectral study

The FTIR pattern of native insulin, blank microspheres fabri-ated by PLGA 504H and 502H and insulin-loaded microspheresf formulations Nos. 1 and 5 were examined and compared (Fig. 6).he spectrum of insulin-loaded microspheres showed two distincteaks at wave numbers 1655 and around 1540 cm−1 that corre-pond to amide regions I and II, respectively. There was no changen the position of peak for amide I region, although peak positionor amide II region shifted toward higher wave number from 1534unprocessed insulin) to 1542 and 1544 for formulations Nos. 5nd 1, respectively. The amide I region (1700–1600 cm−1) containsainly C O stretching vibrations of the polypeptide backbone. The

mide II region (1600–1500 cm−1) contains essentially the contri-ution from C–N stretching and N–H bending modes [45]. Thesebservations suggested that an interaction between insulin andLGA took place during the fabrication of the microspheres. It maye attributed to hydrogenic interaction of N in amide II regionith carboxylic group of PLGA. This is in agreement with studies

eported interactions between negatively charged PLGA and pos-tively charged proteins such as insulin [21,22,31,46,47]. The shifto higher wave length for formulation No.1 was slightly more thanhat of formulation No. 5 which might be attributed to the extrancapped final carboxyl groups of 502H rather than 504H.

.7. DSC study

Fig. 7 depicted the DSC thermograms of insulin, blank micropar-icles of PLGA 504H and PLGA 502H, as well as various drug-loadedarticulate formulations. The thermogram of insulin showed twondothermic peaks at 61.49 and 91.08 ◦C. This thermogram patternf insulin is in agreement with other reported ones in litera-ure [48,49]. The result of detectable Tm peak of insulin in allypes of insulin-loaded microparticles indicated that the drug isntrapped in a crystalline state in the microparticles during sol-ent evaporation process. It was reported that crystalline proteinsre usually less prone to chemical degradation than the amorphousorm [50–52]. It is well known that protein crystallization resultsn the formation of a lattice structure with the protein molecule.irect interaction between protein molecules is relatively weak inost crystals, thus minimizing the loss of biological activity [53].lass transition temperature, Tg, is an important parameter deter-ining the physical strength of the polymeric delivery systems.

he Tg of PLGA 502H and 504H were at about 52.01 and 55.71 ◦C,espectively, which is above physiological temperature, providingufficient strength for drug delivery systems. The higher rate ofnsulin release from PLGA 502H than 504H which was observedn this study could be also attributed to the lower Tg of the 502Hather than 504H polymers. Aso et al. has correlated the Tg of PLAicrospheres to drug release rate. They have shown that higher

g led to lower progesterone release rate [54]. Fig. 7 showed that

m of insulin shifted to higher temperature in all formulationshich indicates insulin-PLGA interaction. Tm of insulin shifted toigher degree after formulation with the low molecular weightLGA (502H) (formulations 1, 112.58 ◦C and 2, 112.08 ◦C) than theigh molecular weight (504H) polymer (formulations 5, 104.40 ◦C

Fig. 8. Wide angle XRD profiles of human insulin, extracted insulin from PLGAmicrospheres and blank microspheres as a function of scattering angle 10◦ < 2� < 70◦ .

and 6, 105.05 ◦C). It may be due to higher degree of ionic interactionof insulin with 502H than 504H which was also confirmed by FTIR(Fig. 6). Comparison of DSC thermograms of formulations 1 with 2,and 5 with 6 revealed that changing in surfactant and polymer con-centrations in the range of this study did not cause any significantchange in Tm of the protein or Tg of the polymer (Fig. 7). In all formu-lations, incorporation of insulin reduced Tg of the polymer whichmay indicate the plasticizing effect of insulin due to its hinderingeffect on the structure of polymer which increases the mobility ofpolymer chains. This fact might be also attributed to the presence ofacetonitrile or hexane but our investigation by GC–mass confirmedthat concentration of residual solvents were less than 0.1 ppm.

3.8. XRD study

Fig. 8 shows the wide angle X-ray scattering profiles ofunprocessed insulin as a reference, extracted insulin from PLGAmicrospheres and blank microspheres. PLGA and insulin in natureis amorphous (Fig. 8) while the extracted insulin powder diffractionpattern shown in Fig. 8 displayed partial sharp crystalline peaks,which is the characteristic of a macromolecule with some crys-tallinity. This is in agreement with DSC results (Fig. 7).

4. Conclusion

The developed single phase o/o emulsion–solvent evaporationmethod allowed the preparation of spherical biodegradable PLGAmicrospheres containing insulin as a model protein and hydrophilicdrug. We have demonstrated that encapsulation efficiency, ini-tial burst, size and size distribution of insulin-loaded microspherescould be controlled by varying preparation conditions such as con-centration of the surfactant and polymer, stirring speed, and thepolymer molecular weight. Fabricated microspheres using PLGAwith higher molecular weight, high surfactant and polymer con-centrations led to a more appropriate encapsulation efficiency ofinsulin with low burst effect and desirable release pattern. FTIR andDSC results confirmed possible drug–polymer interaction that wasconsidered as a reason for higher encapsulation and lower burstrelease of insulin from PLGA 504H than PLGA 502H.

Acknowledgments

for his valuable support to this work (Iran Polymer and Petrochem-ical Institute). They acknowledge Dr. Farzandi and Dr. Attaran forthe generous supply of insulin (Exir Pharmaceutical Co. Iran). Thisproject is financially supported by the Isfahan Medical SciencesUniversity (Grant No. 185104).

urface

R

[[[[[[

[[

[[[

[[

[[[[[[

[

[

[

[[[

[

[[[[

[[[

[

[

[

[[

[

[

H. Hamishehkar et al. / Colloids and S

eferences

[1] A. Vila, A. Sanchez, M. Tobio, P. Calvo, M.J. Alonso, J. Control. Release 78 (2002)15.

[2] S. Cohen, L. Chen, R.N. Apte, React. Polym. 25 (1995) 177.[3] C. Dai, B. Wang, H. Zhao, Colloid Surf. B 41 (2005) 117.[4] B. Kim, N.A. Peppas, Int. J. Pharm. 266 (2003) 29.[5] H. Yushu, S. Venkatraman, J. Appl. Polym. Sci. 101 (2006) 3053.[6] R. Narayani, K.P. Rao, Int. J. Pharm. 143 (1996) 255.[7] A.M. Le Ray, S. Chiffoleau, P. Iooss, G. Grimandi, A. Gouyette, G. Daculsi, C. Merle,

Biomaterials 24 (2003) 443.[8] R.A. Jain, C.T. Rhodes, A.M. Railkar, A.W. Malick, N.H. Shah, Eur. J. Pharm. Bio-

pharm. 50 (2000) 257.[9] J. Herrman, R. Bodmeier, Int. J. Pharm. 126 (1995) 129.10] R. Langer, Acc. Chem. Res. 33 (2000) 94.11] G. Zhu, S.R. Mallery, S.P. Schwendeman, Nat. Biotechnol. 18 (2000) 52.12] J.M. Ruiz, B. Tissier, J.P. Benoit, Int. J. Pharm. 49 (1989) 69.13] R. Bodmeier, H. Chen, Pharm. Pharmacol. 40 (1988) 754.14] J. Herrmann, R. Bodmeier, Eur. J. Pharm. Biopharm. 45 (1998) 75.15] J. Emami, H. Hamishehkar, A. Rouholamini Najafabadi, K. Gilani, M. Minaiyan,

H. Mahdavi, A. Nokhodchi, J. Pharm. Sci. 98 (2009) 1712.16] S.A. Agnihotri, T.M. Aminabhavi, Int. J. Pharm. 324 (2006) 103.17] J. Emami, H. Hamishehkar, A. Rouholamini Najafabadi, K. Gilani, M. Minaiyan,

H. Mahdavi, H. Mirzadeh, A. Fakhari, A. Nokhodchi, J. Microencapsul. 26 (2009)1.

18] J. Fiegel, J. Fu, J. Hanes, J. Control. Release 96 (2004) 411.19] F. Boury, H. Marchais, J.E. Proust, J.P. Benoit, J. Control. Release 45 (1997) 75.20] K. Makino, T. Nakajima, M. Shikamura, F. Ito, S. Ando, C. Kochi, H. Inagawa, G.I.

Soma, H. Terada, Colloid Surf. B 36 (2004) 35.21] D. Blanco, M.J. Alonso, Eur. J. Pharm. Biopharm. 45 (1998) 285.22] M.M. Gaspar, D. Blanco, M.E.M. Cruz, M.J. Alonso, J. Control. Release 52 (1998)

53.23] Y.Y. Yang, T.S. Chung, N.P. Ng, Biomaterials 22 (2001) 231.

24] S. Freitas, H.P. Merkle, B. Gander, J. Control. Release 102 (2005) 313.25] Y.Y. Yang, H.H. Chia, T.S. Chung, J. Control. Release 69 (2000) 81.26] R.C. Mehta, B.C. Thanoo, P.P. DeLuca, J. Control. Release 41 (1996) 249.27] G. Crotts, H. Sah, T.G. Park, J. Control. Release 47 (1997) 101.28] B. Bittner, M. Morlock, H. Koll, G. Winter, T. Kissel, Eur. J. Pharm. Biopharm. 45

(1998) 295.

[[[

[[

s B: Biointerfaces 74 (2009) 340–349 349

29] C. Carrascosa, L. Espejo, S. Torrado, J.J. Torrado, J. Biomater. Appl. 18 (2003)95.

30] J. Rojas, H. Pinto-Alphandary, E. Leo, S. Pecquet, P. Couvreur, E. Fattal, Int. J.Pharm. 183 (1999) 67.

31] G. Jiang, B.H. Woo, F. Kang, J. Singh, P.P. DeLuca, J. Control. Release 79 (2002)137.

32] M.J. Alonso, R.K. Gupta, C. Min, G.R. Siber, R. Langer, Vaccine 12 (1994) 299.33] N.K. Varde, D.W. Pack, Expert Opin. Biol. Ther. 4 (2004) 35.34] F. Cui, K. Shi, L. Zhang, A. Tao, Y. Kawashima, J. Control. Release 114 (2006)

242.35] W.T. Leach, D.T. Simpson, T.N. Val, Z. Yu, K.T. Lim, E.J. Park, R.O. Williams, K.P.

Johnston, AAPS Pharm. Sci. Technol. 6 (2005) E605.36] J. Vandervoort, A. Ludwig, Int. J. Pharm. 238 (2002) 77.37] G.H. Ma, M. Nagai, S. Omi, Colloid Surf. A 153 (1999) 383.38] S. Freiberg, X.X. Zhu, Int. J. Pharm. 282 (2004) 1.39] J.A. Namur, E.C. Cabral-Albuquerque, W. Quintilio, M.H. Santana, M.J. Politi, P.S.

de Araujo, A.C. Lopes, M.H. da Costa, J. Nanosci. Nanotechnol. 6 (2006) 2403.40] P. Sansdrap, A.J. Moës, Int. J. Pharm. 98 (1993) 157.41] F. Gabor, B. Ertl, M. Wirth, R. Mallinger, J. Microencapsul. 16 (1999) 1.42] H. Marchais, F. Boury, C. Damgé, J.E. Proust, J.P. Benoit, S.T.P. Pharm. Sci. 6 (1996)

417.43] R. Jeyanthi, R.C. Mehta, B.C. Thanoo, P.P. DeLuca, J. Microencapsul. 14 (1997)

163.44] N.B. Viswanathan, P.A. Thomas, J.K. Pandit, M.G. Kulkarni, R.A. Mashelkar, J.

Control. Release 58 (1999) 9.45] S. Gélinas, C. Chapados, M. Beauregard, I. Gosselin, M.G. Martinoli, Biochem.

Cell Biol. 78 (2000) 667.46] T.G. Park, H.Y. Lee, Y.S. Nam, J. Control. Release 55 (1998) 181.47] G. Coppi, V. Iannuccelli, E. Leo, M.T. Bernabei, R. Cameroni, J. Microencapsul. 19

(2002) 37.48] K. Huus, S. Havelund, H.B. Olsen, M. van de Weert, S. Frokjaer, Pharm. Res. 23

(2006) 2611.49] B. Sarmento, D. Ferreira, F. Veiga, A. Ribeiro, Carbohydr. Polym. 66 (2006) 1.

50] M.C. Lai, E.M. Topp, J. Pharm. Sci. 88 (1999) 489.51] W. Wang, Int. J. Pharm. 185 (1999) 129.52] Y. Yamaguchi, M. Takenaga, A. Kitagawa, Y. Ogawa, Y. Mizushima, R. Igarashi,

J. Control. Release 81 (2002) 235.53] S. Zhou, X. Deng, X. Li, J. Control. Release 75 (2001) 27.54] Y. Aso, S. Yoshioka, A.L. Wan Po, T. Terao, J. Control. Release 31 (1994) 33.