RESEARCH ARTICLE

Development of sustained-release formulations processedby hot-melt extrusion by using a quality-by-design approach

Muhammad T. Islam & Mohammed Maniruzzaman &

Sheelagh A. Halsey & Babur Z. Chowdhry &

Dennis Douroumis

Published online: 2 April 2014# Controlled Release Society 2014

Abstract In this study, a quality-by-design (QbD) approachwas used to optimize the development of paracetamol(PMOL) sustained-release formulations manufactured byhot-melt extrusion (HME). For the purpose of the study, in-line near-infrared (NIR) spectroscopy as a process analyticaltechnology (PAT) was explored while a design of experiment(DoE) was implemented to assess the effect of the processcritical parameters and to identify the critical quality attributes(CQA) of the extrusion processing. Blends of paracetamol,ethyl cellulose (EC) and Compritol® 888 ATO (C888) wereprocessed using a twin screw extruder to investigate the effectof screw speed, feed rate and drug loading on the dissolutionrates and particle size distribution. The principal componentanalysis (PCA) of the NIR collected signal revealed the opti-mum extrusion processing parameters. Furthermore, the inte-gration of the DoE experiments demonstrated that drug load-ing has a significant effect on the only quality attribute, whichwas the PMOL dissolution rate. This QbD approach wasemployed as a paradigm for the development of pharmaceu-tical formulations via HME processing

Keywords Hot-melt extrusion . In-line NIR . Processanalytical technology . Quality by design . Critical qualityattributes

Introduction

Hot-melt extrusion (HME) has been introduced as an emerg-ing processing technology for the development of variousdrug delivery systems and is expected to significantly affectthe pharmaceutical research and development sector [1].Being a versatile technique, HME not only combines theadvantages of a solvent-free operation with that of a dust-free processing environment [2] but also operates as a contin-uous process for the development of multiple drug deliverysystems. HME has been successfully used not only for taste-masking purposes of various bitter drugs [3] but also toincrease the dissolution rate and bioavailability of poorlywater-soluble drugs via the formation of the solid solutionsof drugs in polymeric matrices during the extrusion process[4, 5]. Furthermore, HME has been used for the production ofextrudates with sustained-release properties. Early studiesthrough HME processing have described the preparation ofmatrix mini tablets which was followed by further investiga-tions into the properties of sustained-release mini matricesmanufactured from ethyl cellulose, HPMC and ibuprofen[6]. Extruded mini tablets showed minimized risk of dosedumping and reduced inter- and intra-subject variability.Recently, vegetable calcium stearate was reported for thedevelopment of retarded-release pellets using a thermoplasticexcipient processed through HME, where pellets with a para-cetamol loading of 20 % showed a release of only 11.54 % ofthe drug after 8 h due to the significant densification of thepellets. As expected, the drug release was influenced by thepellet size and the drug loading [7]. Recently, Almeida et al.has reported an oral sustained-release formulation producedvia HME using ethylene vinyl acetate (EVA) as a potentialpolymer [8].

With the advent of rapid developments of solid dispersions,pharmaceutical industries are moving towards vast changes inthe product and process development bypassing the so-called

M. T. Islam :M. Maniruzzaman : B. Z. Chowdhry :D. Douroumis (*)Faculty of Engineering and Science, University of Greenwich,Medway Campus, Chatham Maritime,Chatham, Kent ME4 4TB, UKe-mail: D.Douroumis@gre.ac.uk

S. A. HalseyThermoFisher Scientific, Stafford House, Boundary Way, HemelHempstead HP2 7GE, UK

Drug Deliv. and Transl. Res. (2014) 4:377–387DOI 10.1007/s13346-014-0197-8

conventional univariate trial approach as well as the risk-based development process. The new approach that is beinghighly encouraged by the regulatory authorities to haveadopted is known as “quality by design” (QbD) which isdescribed in International Conference on Harmonization(ICH) guidelines in various documents such as ICH Q8(R2), ICH Q9 and ICH 10 [9–11]. The main objectives ofQbD approaches are to focus on science-based design anddevelopment of formulation and manufacturing processes inorder to ensure predefined product quality objects [12]. TheQbD approach has already been established in pharmaceuticalresearch to maintain product quality as well as to developdesign space. Recently, Fonteyne et al. used the QbDapproach to predict the quality attributes of continuouslyproduced granules using complementary process analyticaltechnology (PAT) [13]. A Raman and near-infrared (NIR)spectrometer were used together with a photometric imagingtechnique to acquire solid-state information and granule sizedata. The acquired multivariate data were then used to predictthe granules’ moisture content, tapped and bulk density andflowability. QbD approaches have been used in monitoringand mapping the state of a pharmaceutical co-precipitationprocess characterization and design space development [14].It has been employed to examine the effects of testing param-eters and formulation variables on the segregation tendency ofpharmaceutical powder measurements, enhancement of thesolubility and dissolution of class II BCS drugs [15, 16].

To ensure the product quality and in-line measurement ofcritical product parameters, the Food andDrugAdministration(FDA) introduced a new initiative known as process analyticaltechnology (PAT), which is normally used to control andunderstand the manufacturing process [17]. NIR spectroscopyis one of the most common techniques suitable for an in-creased number of PAT applications and has been used inseveral studies in the pharmaceutical and nutritional fields.NIR spectroscopy enables quantitative (e.g. concentrationdetermination) and qualitative analysis including determina-tion of drug crystallinity and identification of polymer–drug orpolymer–polymer interactions via the HME process [18–20].Kelly et al. used NIR as a PAT tool to monitor co-crystalformation during the solvent-free continuous crystallizationprocess [21]. NIR spectroscopy has also been used to testblend uniformity, content uniformity, coating thickness andhardness and quality attributes of continuously produced gran-ules [14] and, more importantly, for upscaling hot-melt-extruded formulations [22]. The main advantage of NIR spec-troscopy is its non-destructive nature and the real-time mon-itoring (immediate delivery of results). NIR is an exclusivetechnique which analyses the samples keeping the main in-strument remote to the sample point. This analysis is furtherfacilitated by the use of a fibre-optic cable connected with theprobe, installed in the process to the NIR instrument. Hot-meltextrusion can be used as a continuous manufacturing

technique where screw speed, feeding rate, temperature pro-file and screw configurations are the critical processing pa-rameters. These parameters affect several properties of theextrudates such as crystallinity, solubility, dissolution ratesand particle size. These are considered as critical qualityattributes. Considering critical quality attributes and criticalprocessing parameters, a QbD approach can be integrated inhot-melt extrusion to ensure the final product quality anddevelop a design space.

The article describes the application of NIR spectroscopyas a PAT tool during the hot-melt extrusion process by using aQbD approach combined with a design of experiment (DoE).NIRwas used for in-line monitoring of the effect of the criticalprocessing parameters such as screw speed and drug loadingduring extrusion in order to identify the best processing con-ditions of paracetamol (PMOL) in polymer/lipid matrices.Drug loading was included in the DoE experiment, and theobtained data were analysed to identify the significant inputvariables. The dissolution profiles of the sustained-releaseformulations were selected as independent variable.

Materials and methods

Materials

PMOL, ethyl cellulose (EC) and Compritol 888 ATO (C-888)were kindly donated by Mallinckrodt Chemical Ltd.(Canada), Colorcon (Dartford, UK) and Gattefosse (Saint-Priest, France), respectively. The HPLC solvents were ofanalytical grade and purchased from Fisher Chemicals (UK).All other materials were used as received.

Hot-melt extrusion

Hot-melt extrusion was performed by using a Eurolab 16-mmtwin screw extruder (Thermo Fisher Scientific, Germany).The hot-melt extruder was equipped with a DD Flexwall®18 feeder (Brabender Technology, Germany), which was setin its gravimetric feeding mode. Three different drug/lipid/polymer blend ratios were extruded, consisting of equalamount of EC and Compritol-888 ATO (C888) with 40, 60and 80 % (w/w) PMOL, respectively. Prior to hot-melt extru-sion, the drug, polymer and lipid were blended in a Turbulamixer (100 rpm) (Basel, Switzerland) to mix them homoge-neously for 10 min each. The barrel temperature profile wasset at 50–100–140–140–140–140–140–140–100 °C (fromfeeder to die) for all batches. Extrudates with different pro-cessing parameters were obtained by incorporating a three-factor response surface fraction factorial design (DoE) inrandomized order by using Fusion One software (DoEFusion OneTM, USA). Three centrepoint experiments wereperformed as well, resulting in 23+3=11 experiments. Drug

378 Drug Deliv. and Transl. Res. (2014) 4:377–387

loading (% w/w), screw speed and feed rate (kg/h) wereconsidered as variables whereas drug release rate was mea-sured as dependant variables/response.

NIR spectroscopy

Diffuse reflectance NIR spectra were continuously collectedin-line and non-invasively during hot-melt extrusion using aFourier-transform NIR spectrometer (Thermo FisherScientific, UK, Antaris MX near-IR analyser) equipped withan InGaAs detector, a quartz halogen lamp and a fibre-opticprobe which was attached in the extrusion die. Spectra werecollected every 1 min in the 1,000–4,000-cm−1 region with aresolution of 16 cm−1 and averaged scan over 32 scans.

Data analysis was performed by using the Result Software(version 3.0, Thermo Fisher Scientific, UK). Spectra collectedin the diffuse reflectance mode usually require spectral pre-treatment before analysis. The degree of scattering depends onthe wavelength of the light and the refractive index of thesample, which causes a no-equal scatter over the whole spec-trum. This can result in a baseline shift. Therefore, standardnormal variate (SNV) correction was used before chemomet-ric analysis of the spectra. Using SNV, unwanted scatter wasremoved from the raw spectra to prevent it from dominatingover chemical information within the spectra. The result ofSNV pre-processing is that each spectrum has the same offsetand amplitude, eliminating difference in light scatter in thespectra from different samples, before developing the calibra-tion model. Furthermore, second derivative pre-processingwas done after SNV correction. Second derivatives of NIRspectra magnify differences in spectral features, provide base-line normalization and remove data offsets due to scattering

effects and path length variation. For principal componentanalysis (PCA) and for the development of the partial leastsquare (PLS) model, five spectra of each drug–polymer–lipidmixture were used. Prior to PCA and PLS, spectra were meancentred. The PLS model was developed by regressing thePMOL concentration (Y) versus the corresponding in-linecollected spectra (X). This model was validated by usingnew NIR spectra collected of each drug–polymer–lipidmixture. These validation spectra were used to evaluate thepredictive performance of the PLS model.

Thermal analysis

The physical states of the pure drug and the extrudedgranules were studied by differential scanning calorimetry(DSC) using a METTLER TOLEDO 823e (Greifensee,Switzerland) differential scanning calorimeter. Samplesaccurately weighed (2–3 mg) were placed in sealed aluminiumpans. Measurements were carried out in a nitrogen atmosphere.The flow rate of dry nitrogen gas was 50 ml/min. Samples wereheated from 0 to 220 °C at 10 oC/min in, kept at 220 oC for1 min and cooled down to 0 oC at 10 oC/min.

X-ray powder diffraction

X-ray powder diffraction (XRPD) was used to assess thecrystalline state of the active substance in the extruded for-mulations. All formulations including pure PMOL, and physi-cal mixtures and extruded formulations were evaluated usinga Bruker D8 Advance in theta–theta mode, Cu anode at 40 kVand 40 mA, parallel beam Goebel mirror, 0.2-mm exit slit,LynxEye position-sensitive detector with 3° opening and

Fig. 1 Thermogram of the bulkPMOL, C888 and EC (n=3)

Drug Deliv. and Transl. Res. (2014) 4:377–387 379

LynxIris at 6.5mm, sample rotation at 15 rpm. The sampleswerescanned from 2 to 40° 2-theta with a step size of 0.02° 2-thetaand a counting time of 0.2 s per step; 176 channels were active onthe PSD making a total counting time of 35.2 s per step.

Tablet preparation

For the tablet compression, all batches were preparedusing a 500-g dose of the API. All excipients were

passed through a mesh sieve with an aperture of500 μm before use. The batches were blended withmagnesium stearate (1 %) and silicon dioxide (0.25 %)in a Turbula TF2 mixer (Basel, Switzerland) for 5 min.Blends were directly compressed on a FlexiTab tabletpress (OYSTAR Manesty, Germany) using 13-mm nor-mal flat punches. Dwell time was set at 30 ms, and thecompaction force varied from 8 to 12 kN to obtain thefinal tablets with a hardness of 5–6 kP.

Fig. 2 Thermograms of theextrudates containing 40, 60 and80 % PMOL with equivalentamounts of EC and C888 (n=3)

Fig. 3 Thermograms of theextrudates containing 40, 60 and80 % PMOL with equivalentamounts of EC and C-888 ATO(n=3)

380 Drug Deliv. and Transl. Res. (2014) 4:377–387

In vitro drug release study

In vitro drug release studies were carried out in 750 mlof 0.1 M hydrochloric acid for 2 h using a Varian 705-DS dissolution paddle apparatus (Varian Inc., NC, USA)at 100 rpm and 37±0.5 °C. After 2 h of operation,150 ml of a 0.20 M solution of trisodium phosphatedodecahydrate was added into each of the vessels (buff-er stage, pH 6.8) that have been equilibrated to 37±0.5 °C. At predetermined time intervals, samples werewithdrawn for HPLC assay. All dissolution studies wereperformed in triplicate.

HPLC analysis

The release of PMOL was determined by an AgilentTechnologies HPLC 1200 series system equipped with aHICROM S50DS2, 5 μm×150 mm×4 mm column. TheUV wavelength used for the analysis was set at 276 nm forthe PMOL HPLC assay. The mobile phase consisted ofacetonitrile/water (1 % acetic acid) (50:50, v/v). The flow ratewas adjusted at 1.5 ml/min, and the retention time of PMOLwas about 4 min. The PMOL calibration curve plotted withthe concentrations varying from 10 to 50 μg/ml was used toevaluate all the samples with 20 μl injection volume.

Fig. 4 Thermograms of thephysical mixture containing 40,60 and 80 % PMOL withequivalent of amounts of EC andC888 (n=3)

Fig. 5 X-ray diffractograms ofthe physical mixture containing80 % PMOL and extrudates withthe same PMOL loadings (n=3)

Drug Deliv. and Transl. Res. (2014) 4:377–387 381

Results and discussion

HME processing

The objective of this study was not only to develop sustained-release PMOL tablets processed by HME but also to investi-gate the effect of the processing parameters which eventuallywill help to identify the critical quality attributes (QCAs) anddefine the design space. For the formulation and processoptimization, a DoE was implemented where the processparameters (screw speed, feed rate) and the drug loading wereused as inputs, while PMOL dissolution rates were the depen-dent variables (response). The relationship and the effect of

the three inputs on the dissolution rate were further explored.In order to reduce experimentation and the number of pro-cessing parameters, we did not include temperature in theDoE, but the optimized temperature profile (data not shown)was used for all experiments. C888 acted as a plasticizerresulting in reduced extrusion temperatures (140 °C) com-pared to PMOL/EC binary blends (>150 °C).

The rationale for the co-processing of lipid–polymer for-mulations is not uncommon and has been reported byWindbergs et al. by using trilaurin/polyethylene glycol(PEG) blends with various drugs to adjust the release profiles(e.g. increased rates) [23, 24]. In our case, C888 was intro-duced in order to reduce the burst release in PMOL/EC

Fig 6 X-ray diffractograms of the physical mixture containing 60 % PMOL and extrudates with the same PMOL loadings (n=3)

Fig. 7 X-ray diffractogram of the physical mixture containing 40 % PMOL and extrudates with the same PMOL loadings (n=3)

382 Drug Deliv. and Transl. Res. (2014) 4:377–387

extrudates (data not shown). All formulations were easilyprocessed even at 80 % PMOL loadings. The particle sizedistribution was measured for all extruded formulations whileCarr’s indexes varied from 5 to 12 %. Despite the fact thatHME processing was conducted at different settings, thecompressed tablets were robust with low friability (<0.1 %),and hardness varied from 5.0 to 6.0 kP.

Thermal analysis

DSCwas used to study the solid state of PMOLwithin the EC/C888 matrices. As can be seen in Fig. 1, the melting peak ofbulk PMOL was observed at 172.41 °C (ΔH=144.40 J g−1)while for C888, at 72.47 °C (ΔH=139.05 J g−1). The DSCthermogram of EC presented a Tg at 117.61 °C due to itsamorphous nature. It has been previously reported [25–27]that PMOL presents three polymorphic forms such as thestable form I (m.p. 170 °C), form III, which is highly unstable(m.p. 148 °C) and the metastable form II (m.p. 160 °C). Basedon the data obtained from the DSC analysis, it can be seen thatthe polymorphic form of PMOL used in the current study wasform I. The DSC scans of the PMOL/EC/C888 extrudates(Figs. 2 and 3) showed melting endotherms at 167, 166 and165 °C corresponding to 80, 60 and 40 % (w/w) PMOLloadings, respectively. The observed melting peaks are shiftedto lower temperatures, and the peak shapes are broader com-pared to those of pure PMOL suggesting the presence of lesscrystalline PMOL after extrusion, but still, the observed melt-ing peaks of PMOL between 165 and 167 °C indicate thepresence of form I in the lipid–polymer matrices. The shift ofthe melting point in all drug/polymer/lipid physical mixtures(Fig. 4) as well as in the extruded formulations is attributed tothe solubilization effect of the lipid/polymer and the smallamorphous PMOL content.

Sweeney et al. showed that partial drug solubilization cantake place due to the slow DSC heating rates which providesufficient time for C888 to solubilize PMOL [28]. However,the broad PMOL melting endotherms suggest the presence of

Fig. 8 NIR spectra of PMOL(bulk), EC, C888 and PMOL/EC/C888 extruded formulations (32scans)

Table 1 Experimental design of PMOL/C888/EC formulations withdrug loading, screw speed and feed rate as independent variables andPMOL release (%) and particle size D(0, 5) as dependent variables

Run no. Independent variables Dependant variables

Drugloading(% w/w)

Screwspeed(rpm)

Feedrate(g/h)

PMOLrelease(T6 h %)

PMOLrelease(T12 h, %)

Particlesize D(0, 5)

F1 80 150 400 61.5 86.89 273.98

F2 60 150 700 47.77 75.61 250.52

F3 40 150 700 38.16 51.32 202.78

F4 80 50 1,000 63.78 90.34 209.44

F5 80 250 700 67.45 92.67 163.61

F6 60 150 1,000 47.22 74.46 606.44

F7 40 250 1,000 41.20 59.27 529.64

F8 40 50 400 35.43 48.35 466.59

F9 60 50 700 50.11 79.35 252.29

F10 60 150 700 46.88 73.41 282.93

F11 60 250 400 48.84 75.45 227.32

Drug Deliv. and Transl. Res. (2014) 4:377–387 383

a small amorphous drug content due to possible PMOL/ECinteractions or the processing conditions. This is obvious fromthe variability of the ΔH values of the processed formulationat the same drug loadings.

X-ray powder diffraction

XRPD was used to investigate the crystalline state of PMOLwithin the lipid–polymer matrices. The standard XRPD pat-terns of pure PMOL, physical mixtures with EC/C-888 ATOand extrudates are depicted in Figs. 5, 6, and 7, respectively.The bulk PMOL showed distinct peaks at 12.09°, 13.78°,15.5°, 18.17°, 20.35°, 23.46°, 24.74° and 26.52° 2θ valuesand a series of smaller peaks ranging from 27.16° to 39.50°

2θ. C888 peaks appeared at 4.13°, 21.14° and 23.33° 2θvalues whereas EC showed diffused peaks indicating theamorphous nature of the polymer. The diffraction patterns ofthe physical mixtures at three different ratios (40, 60 and80 %) presented identical crystalline peaks to those of purePMOL but with relatively lower intensities. The XRPD pat-terns of extruded formulations showed characteristic peakscorresponding to bulk PMOL, which indicates the presenceof crystalline PMOL in the solid dispersion prepared viaHME. Furthermore, the diffractions patterns of the allextrudates confirmed the presence of form I within the poly-mer matrices, and no new distinct crystalline peaks at different2θ were observed [26]. In addition, the peak intensities of theextruded formulations appear smaller than those of the

Fig. 9 PLS calibration curveobtained by 40, 60 and 80 %PMOL with varying processingparameters (32 scans)

Fig. 10 Principal component plotobtained after PCA on the pre-processed in-line collected NIRspectra during extrusion ofmixtures containing 40, 60 and80 % PMOL extruded with ECand C888

384 Drug Deliv. and Transl. Res. (2014) 4:377–387

respective physical mixtures with equal drug loadings sug-gesting a different content of amorphous PMOL. This obser-vation is in agreement with the DSC results and confirmed theco-existence of low amorphous PMOL content. Previousstudies [29] showed that C888 cannot facilitate PMOLamorphicity during HME processing, and thus, it should beattributed to the presence of EC. Indeed, Windbergs et al. [23,24] confirmed that the presence of PEG affected the solid stateof theophylline due to the drug–polymer interactions, whichled to changes of the drug crystalline state (anhydrous). Thesame group showed that the processing parameters (e.g. tem-perature) could significantly affect the drug state and conse-quently the dissolution rates. In our case, EC presented a smallsolubilization capacity leading to a small PMOL amorphous

content, which is also affected by the processing parameters(e.g. screw and feed rates) when the same PMOL loadingswere processed. Nevertheless, stable PMOL solid dispersionswere manufactured with reproducible solid-statecharacteristics.

In-line NIR monitoring

Off-line NIR spectra of PMOL, EC, C888 and the physicalmixture (PM) were measured to identify the characteristicpeak attributed to the pure samples. From Fig. 8, it can beeasily observed that the NIR spectra of pure samples havecharacteristic peaks at different wavelengths.

All extrudates containing 40, 60 and 80 % PMOL (w/w)blends (Table 1) with EC and C888 matrices were monitoredwith the in-line near-infrared probe where the HME process-ing parameters (screw speed and feed rate), drug loading andpercent drug release of the active substance were taken intoaccount as critical processing parameters (CPP) and criticalquality attributes (CQA) of the PAT processing, respectively.The typical in-line NIR spectra collected during extrusionprocessing (9,091–4,085 cm−1) are shown in Fig. 8. A PLScalibration model (Fig. 9) was developed to allow predictionof PMOL concentration in unknown samples during the hot-melt extrusion process (r2=0.99). The in-line NIR spectra thatwere regressed versus the known PMOL concentrations es-tablish the rationale of the foregoing claims.

A PCAwas applied for all 55 NIR spectra (five spectra foreach extruded formulation) leading to a model with twoprincipal components covering nearly all spectral variations.The first principal component (PC1) captures 83.982 % of thevariation while the second component (PC2) covers 7.295 %

Fig. 11 Dissolution profiles ofPMOL/EC/C888 formulations at40, 60 and 80 % drug content,extruded at different processingsettings (n=3)

Table 2 ANOVA analysis for the PMOL tablets release (%) (T6 h, T12 h)and particle size of the extrudates according to the response surface design

Coded variablename

p values of correlation coefficients

PMOL release(6 h)

PMOL release(12 h)

Particle sizeD(0, 5)

X1 0.0239 0.0396 0.1900

X2 0.5660 0.9815 0.3293

X3 0.1788 0.3199 0.1819

(X1)2 0.0921 0.1484 0.1704

(X2)2 0.0926 0.1469 0.5916

(X3)2 0.1683 0.3136 0.0825

X1×X2 0.3979 0.9027 0.2045

X1×X3 0.2372 0.3555 0.2374

X2×X3 0.3418 0.9169 0.5401

X1 drug loading, X2 screw speed, X3 feed rate

Drug Deliv. and Transl. Res. (2014) 4:377–387 385

extra variation (Fig. 10). The PC score plot for spectra with thesame process conditions with three different PMOL loadingsdescribes the changes in PMOL concentration very well. Thesizes of the ellipses for each concentration varied which showsthat possibly there is extra variation attributed with PMOLconcentration. In most of the instances, the ellipses are smallerat high drug loading compared to that of the lower loading,which demonstrates that the low drug loading introduces morevariability into the spectra of the extruded formulations. Forthat reason, all formulations with 80 % PMOL loadings (rightpart of the PCA plot) showed relatively lower variabilitycompared to that of 60 % PMOL-loaded formulations.Similarly, all formulations with 60% PMOL loadings showedvariability at a lesser extent compared to that of 40 % loadingsbut, as expected, slightly higher than 80 %. The same reasonsaccount for the observed smaller variation due to the feedingrate along the PC2.

These data showed the sensitivity of NIR spectroscopy tothe processing conditions during extrusion. Nevertheless, thedata also show the potential implementation and modernapplication of the NIR spectroscopy system to optimize crit-ical process conditions.

In vitro dissolution studies

In vitro dissolution studies were carried out for the com-pressed PMOL tablets at equal doses of 500 mg to assess theirperformance. In Fig. 11, it can be seen that the high drug-loaded formulations (F1, F4 and F5) showed higher dissolu-tion rates with more than 60 and 80 % of the drug beingdissolved after 6 and 12 h, respectively. In contrast, tabletswith 40 and 60 % PMOL loadings showed sustained releasebut at much lower rates at 6 and 12 h. It is obvious that PMOLrelease rates depend on the final drug loading due to the highhydrophilic nature of the drug. These types of release patternshave been recently reported by Haupt et al. [30] where in-creasing loadings of trospium chloride in lipidic matricesshowed different dissolution rates. The combination of EC/C888 matrices could successfully control PMOL release rateswhile it is worth noticing the absence of burst release for allformulations.

The PMOL release percentages at 6 and 12 h were used asindependent variables in the DoE in order to assess the effectof the processing parameters and the drug loading. As shownin Table 2, the ANOVA statistical analysis revealed that onlydrug loading has a significant effect on PMOL dissolutionrates with probability values (p) less than 0.05. None of theprocessing parameters showed a significant effect on bothdissolution rates and particle size distribution.

The PCA analysis showed that only F4, F5 and F6 fulfil thecriteria for HME processing, and the processing parametersintroduce low variability during processing. In contrast, asshown from the DoE analysis, only F1, F4 and F5 provide

the desired PMOL release rates, and sustained release above80 % can be obtained after 12 h. The QbD approach intro-duced in this study, whereas a DEO experimentation and in-line PAT tools are combined, helped to identify a CQA andoptimize the extrusion processing. Eventually, the employedQbD approach helped to identify the design space for thisparticular dosage form. Nevertheless, the same strategy couldbe used in the future for other formulations by introducingadditional independent and dependant variables.

Conclusions

HME was used to develop PMOL/EC/C888 sustained-releaseformulations and to investigate the effect of screw rate, feedrate and drug loading. By implementing a DoE approach, itwas revealed that only drug loading had a significant effect onthe drug dissolution rates. The integration of in-line NIRmonitoring during HME processing helped to identify theoptimum extrusion processing parameters. This paradigm ofQbD approach can be used in the future for process optimi-zation and better understanding of the developed pharmaceu-tical formulations.

Conflict of interest Authors Islam, Maniruzzaman, Hasley, Chowdhryand Douroumis declare no conflict of interest.

References

1. Gavin PA, David SJ, Osama AD, Daniel NM, Mark SMc. Hot-meltextrusion: an emerging drug delivery technology. Pharm TechnolEur. 2009;2(1)

2. Maniruzzaman M, Boateng JS, Snowden MJ, Douroumis D. Areview of hot-melt extrusion: process technology to pharmaceuticalproducts. IRRN Pharm. 2012;2012:436763.

3. Maniruzzaman M, Boateng JS, Bonnefille M, Aranyos A, MitchellJC, Douroumis D. Taste masking of paracetamol by hot-melt extru-sion: an in vitro and in vivo evaluation. Eur J Pharm Biopharm.2012;80(2):433–42.

4. Cilurzo F, Cupone IE, Minghetti P, Selmin F, Montanari L. Fastdissolving film made of maltodextrins. Eur J Pharm Biopharm.2008;70(3):895–900.

5. Kalivoda A, FischbachM, Kleinebudde P. Application of mixtures ofpolymeric carriers for dissolution enhancement of fenofibrate usinghot-melt extrusion. Int J Pharm. 2012;429(1–2):58–68.

6. De Brabander C, Vervaet C, Remon JP. Development and evaluationof sustained release mini-matrices prepared via hot melt extrusion. JControl Release. 2003;89:235–47.

7. Roblegg E, Jäger E, Hodzic A, Koscher G, Mohr S, Zimmer A, et al.Development of sustained-release lipophilic calcium stearate pelletsvia hot melt extrusion. Eur J Pharm Biopharm. 2011;79:635–45.

8. Almeida A, Possemiers S, Boone MN, De Beer T, Quinten T, VanHoorebeke L. Ethylene vinyl acetate as matrix for oral sustainedrelease dosage forms produced via hot-melt extrusion. Eur J PharmBiopharm. 2011;77(2):297–305.

386 Drug Deliv. and Transl. Res. (2014) 4:377–387

9. ICH. International Conference on Harmonisation (ICH) of technicalrequirements for registration of pharmaceuticals for human use,Topic Q8 (R2): Pharmaceutical development, Geneva, 2009.

10. ICH. International Conference on Harmonisation (ICH) of technicalrequirements for registration of pharmaceuticals for human use,Topic Q9: Pharmaceutical quality system, Geneva, 2005.

11. International Conference on Harmonisation (ICH) of technical re-quirements for registration of pharmaceuticals for human use, TopicQ10: Pharmaceutical quality system, Geneva, 2008.

12. Yu LX. Pharmaceutical quality by design: product and process de-velopment, understanding, and control. Pharm Res. 2008;25(4):781–91.

13. Fonteyne M, Soares S, Vercruysse J, Peeters E, Burggraeve A,Vervaet C, et al. Prediction of quality attributes of continuouslyproduced granules using complementary PAT tools. Eur J PharmBiopharm. 2012;82(2):429–36.

14. Wu H, White M, Khan MA. Quality-by-design (QbD): an integratedprocess analytical technology (PAT) approach for a dynamic phar-maceutical co-precipitation process characterization and process de-sign space development. Int J Pharm. 2011;405(1–2):63–78.

15. Adam S, Suzzi D, Radeke C, Khinast JG. An integrated quality bydesign (QbD) approach towards design space definition of a blendingunit operation by discrete element method (DEM) simulation. Eur JPharm Sci. 2011;42(1–2):106–15.

16. Xie L, Wu H, Shen M, Augsburger LL, Lyon RC, Khan MA, et al.Quality-by-design (QbD): effects of testing parameters and formula-tion variables on the segregation tendency of pharmaceutical powdermeasured by the ASTM D 6940-04 segregation tester. J Pharm Sci.2008;97(10):4485–97.

17. Food and Drug Administration CDER. Guidance for industry, PAT—a framework for innovative pharmaceutical development,manufacturing, and quality assurance. 2006

18. Saerens L, Dierickx L, Lenain B, Vervaet C, Remon JP, DeBeer T. Raman spectroscopy for the in-line polymer–drugquantification and solid state characterization during a phar-maceutical hot-melt extrusion process. Eur J Pharm Biopharm.2011;77(1):158–63.

19. Saerens L, Dierickx L, Quinten T, Adriaensens P, Carleer R, VervaetC. In-line NIR spectroscopy for the understanding of polymer–drug

interaction during pharmaceutical hot-melt extrusion. Eur J PharmBiopharm. 2012;81(1):230–7.

20. De Beer T, Burggraeve A, Fonteyne M, Saerens L, Remon JP,Vervaet C. Near infrared and raman spectroscopy for the in-processmonitoring of pharmaceutical production processes. Int J Pharm.2011;417(1–2):32–47.

21. KellyAL, Gough T, Dhumal RS, Halsey SA, Paradkar A.Monitoringibuprofen–nicotinamide cocrystal formation during solvent free con-tinuous cocrystallization (SFCC) using near infrared spectroscopy asa PAT tool. Int J Pharm. 2012;426(1–2):15–20.

22. Almeida A, Saerens L, De Beer T, Remon JP, Vervaet C. Up scalingand in-line process monitoring via spectroscopic techniques of eth-ylene vinyl acetate hot-melt extruded formulations. Int J Pharm.2012;439(1–2):223–9.

23. Windbergs M, Strachan CJ, Kleinebudde P. Understanding the solid-state behaviour of triglyceride solid lipid extrudates and its influenceon dissolution. Eur J Pharm Biopharm. 2009;71(1):80–7.

24. Windbergs M, Strachan CJ, Kleinebudde P. Influence of structuralvariations on drug release from lipid/polyethylene glycol matrices.Eur J Pharm Biopharm. 2009;37(5):555–62.

25. Rossi A, Savioli A, Bini M, Capsoni D, Massarotti V, Bettini R, et al.Solid-state characterization of paracetamol metastable polymorphsformed in binary mixtures with hydroxypropylmethylcellulose.Thermochim Acta. 2001;406:55–67.

26. Qi S, Gryczke A, Belton P, Craig DQM. Characterisation of soliddispersions of paracetamol and EUDRAGIT® E prepared by hot-melt extrusion using thermal, microthermal and spectroscopic analy-sis. Int J Pharm. 2008;354:158–67.

27. Qi S, Avalle P, Saklatvala R, Craig DQM. An investigation into theeffects of thermal history on the crystallisation behaviour of amor-phous paracetamol. Eur J Pharm Biopharm. 2008;69:364–71.

28. Sweeney G. (thesis) Queen’s Belfast University 2012.29. Vithani K, Maniruzzaman M, Slipper IJ, Mostafa S, Miolane C,

Cuppok Y, et al. Sustained release solid lipid matrices processed byhot-melt extrusion (HME). Colloids Surf B: Biointerfaces. 2013;110:403–10.

30. Haupt M, Thommes M, Heidenreich A, Breitkreutz J. Lipid-basedintravesical drug delivery systems with controlled release of trospiumchloride for the urinary bladder. J Control Rel. 2013;170:161–6.

Drug Deliv. and Transl. Res. (2014) 4:377–387 387

本文献由“学霸图书馆-文献云下载”收集自网络，仅供学习交流使用。

学霸图书馆（www.xuebalib.com）是一个“整合众多图书馆数据库资源，

提供一站式文献检索和下载服务”的24 小时在线不限IP

图书馆。

图书馆致力于便利、促进学习与科研，提供最强文献下载服务。

图书馆导航：

图书馆首页 文献云下载 图书馆入口 外文数据库大全 疑难文献辅助工具