Abstract - KU Leuvenet+al_IJP_…  · Web viewThis leads to their continuous impregnation and drying thus resulting in drug-loaded ... single processing unit procedure was applied

Drug loaded and ethylcellulose coated mesoporous silica for controlled drug release

prepared using a pilot scale fluid bed system

Youcef Chakib Hacene, Abhishek Singh, Guy Van den Mooter*

Drug Delivery and Disposition, KU Leuven, Leuven, Belgium

*Corresponding author:

Address- Drug Delivery and Disposition, Department of Pharmaceutical and

Pharmacological Sciences, University of Leuven; Campus Gasthuisberg O+N2;

Herestraat 49 b921, 3000 Leuven; BELGIUM

Tel.: +32 16 330 304 fax: +32 16 330 305 Mobile: +32 473 356 132

e-mail: [email protected]

1

Abstract

The goal of this study was to test the feasibility to load non-ordered, non-

sperical mesoporous silica with the model drug paracetamol, and subsequently coat

the loaded particles using one single pilot scale fluid bed system equipped with a

Wurster insert. Mesoporous silica particles (Davisil®) with a size ranging from 310 to

500 µm and an average pore diameter of 15 nm were loaded with paracetamol to

18.8% drug content. Subsequently, loaded cores were coated with ethylcellulose to

obtain controlled drug release. Coating processing variables were varied following a

full factorial design and their effect on drug release was assessed. Increasing coating

solution feed rate and decreasing fluidizing air temperature were found to increase

drug release rates. Increasing pore former level and decreasing coating level were

found to increase drug release rates. The release medium’s osmolality was varied

using different sodium chloride concentrations, which was found to affect drug

release rates. The results of this study clearly indicate the potential of non-ordered,

non-spherical mesoporous silica as a reservoir carrier for the controlled release of

drugs. Although non-spherical, we were able to reproducibly coat this carrier using a

bottom spray fluid bed system. However, a major hurdle that needs to be tackled is

the attrition the material suffers from during fluid bed processing.

Keywords

Mesoporous silica, controlled drug release, coating, drug loading, ethylcellulose, PVP

K30

2

1. Introduction

The oral route is still the preferred option for drug administration due to its minimal

invasiveness and requirement for assistance. Controlled oral drug delivery presents

several advantages over immediate release, such as increased patient compliance,

reduced side effects and maintenance of the plasma drug concentrations for longer

periods of time ( Conley et al., 2006; Salsa et al., 1997). This mode of drug delivery

has been extensively studied and as a result several techniques and technologies

have been developed. The matrix system is the most widely used and typically

consists of a carrier and active pharmaceutical ingredient (API) mixture from which

the API is released in a predetermined manner (Tiwari and Rajabi-Siahboomi, 2008).

The reservoir system has a core-shell structure, with a drug containing core coated

with a polymeric membrane (Tang et al., 2012). When the coating is water insoluble it

controls drug release which can be prolonged over a period of time. From such

systems, drug molecules are released into the aqueous medium via three principal

mechanisms viz. drug diffusion through the continuous polymer phase, drug diffusion

through water filled pores situated within the coating membrane, and osmotic

pumping when an osmotic pressure gradient between the inside and the outside of

the reservoir exists (Ozturk et al., 1990). Multiple steps are involved in the

preparation of the reservoir systems starting with the preparation of the drug

containing cores which are typically either granules or extrudates, or layered inert

pellets. Next a polymeric film is applied on the cores containing the drug using a pilot

scale fluid bed system (Felton and Porter, 2013; Guignon et al., 2002).

According to the IUPAC nomenclature, mesoporous materials are porous

structures with a pore diameter ranging between 2 to 50 nm. Mesoporous silica is

made up of silica-based porous frameworks which are characterized by a wide

3

surface area that permits the adsorption of the guest molecules ( Singh et al., 2011).

With respect to oral drug delivery, these materials have been widely studied as drug

carriers during the past decade by various research groups for either controlled drug

delivery or for improving bioavailability of poorly water soluble drugs (Mellaerts et al.

2008; Limnell et al., 2011; Vallet-Regi et al., 2001; Vialpando et al., 2011). It has

been shown that release of drug molecules from such carriers can be prolonged by

tuning the pore diameter (Qu et al., 2006), pore length and surface chemistry of the

carrier (Song et al., 2005), or delayed by functionalization with pH-responsive ligands

(Pérez-Esteve et al., 2015).

Various loading methods for mesoporous silica materials have been reported

(Limnell et al., 2011; Mellaerts et al., 2008). The most common approach consists of

soaking the porous materials with a drug containing solution followed by solvent

evaporation. A practical way to achieve such a process is by using a fluid bed

system, where the porous carrier particles are suspended in a drug solution and

spray-dried within the fluid bed reactor (Limnell et al., 2011) or fluidized and sprayed

with a drug containing solution (Nagane et al., 2014). This leads to their continuous

impregnation and drying thus resulting in drug-loaded porous carrier particles.

The objective of this study was to investigate the potential/feasibility of non-

ordered, non-spherical mesoporous silica to be processed in a pilot scale fluid bed

system as a reservoir carrier for controlled drug release of poorly soluble drugs.

Coating of mesoporous silica particles is very challenging due to its non-spherical

nature. The coating was performed by a bottom spraying fluid bed system equipped

with a Wurster insert. Paracetamol was chosen as a model drug and ethylcellulose

as model coating polymer. A multiple step, single processing unit procedure was

applied in order to load the porous particles with the drug, coat the drug-loaded

4

particles with a polymeric rate controlling membrane, and dry off the residual solvent.

Three processing parameters, namely fluidizing air volume, coating solution feed rate

and fluidizing air inlet temperature, were varied during the coating step according to a

full factorial design. A total of 9 experiments were performed in order to assess the

effect of processing conditions on drug release characteristics. Out of the 9

experiments conducted, experimental conditions of one of these experiments was

chosen and kept constant for further investigations and for which the experimental

variance was investigated. The quantitative and qualitative composition of the coating

layer’s effect on the product’s characteristics was examined. Particle size distribution,

scanning electron microscopy (SEM) and the effect of release medium osmolality on

drug release were also determined.

5

2. Materials and methods

2.1. Materials

Davisil® LC150A was purchased from Grace (Worms, Germany). These

mesoporous silica particles are characterized by non-sphericity, a mean pore

diameter of 15 nm, total surface area of 330 m²/g and particle size ranging from 315

to 500 µm. Paracetamol was purchased from Fagron (Saint Denis, France).

Ethylcellulose (48-49.5% ethoxyl content) was acquired from Sigma Aldrich (The

Netherlands). Polyvinylpyrrolidone K30 (PVP K30) was a gift sample from BASF

(Ludwigshafen, Germany). Dibutyl Sebacate was obtained from TCI chemicals

(Tokyo, Japan), Disinfectol®, a 95:5 v:v ethanol:ether mixture was purchased from

Chem-Lab nv (Zedelgem, Belgium). Sodium Chloride was acquired from Fisher

Chemicals (Leicestershire, United Kingdom), Ultrapure water was produced from an

ELGA Maxima apparatus (Bucks, England).

2.2. Methods

2.2.1. Carrier loading and coating

The overall process of loading and coating of the mesoporous cores can be

divided into 5 steps. In the first step, the carrier was loaded with a paracetamol

solution. An Aeromatic MP1 multiprocessor (GEA Pharma, Bubendorf, Switzerland)

fluid bed system was employed. 400 grams of the carrier was fluidized and sprayed

with a 10% w/v solution (obtained by adding 100 g of ultrapure water to 100 g of

paracetamol, and further making up the volume to 1 liter with denatured ethanol).

The fluidizing air rate (expressed as function of the pressure drop between the inlet

and outlet air) was 25 Pa, spraying air pressure 1 bar, solution feed rate 28 ml/min,

inlet air temperature 35 °C, and a total spray duration of 35 minutes were used as

process parameters. In the second step, solvent evaporation was further continued

6

by drying the particles during 25 min in the fluid bed system using the same

parameters.

Due to attrition of the silica particles during the fluidization process, we had to

build in a sieving step. In this step the loaded particles were sieved to obtain particles

with size >300 µm. Subsequently, the material was coated with a 9% w:v solution of

ethylcellulose (always containing dibutyl sebacate as plasticizer in a concentration of

25% wrt ethylcellulose). The spray solution was heated at 35 °C and stirred during

the entire coating operation. The final step consisted of drying the coated particles in

situ for 30 minutes at 50 °C with an air volume fixed at 10 Pa and the atomizing air

pressure switched off.

2.2.2. Experimental design

In order to study the effect of process parameters on the drug release properties,

and to select the most appropriate experimental conditions, a full factorial design with

a central point (a total of 9 experiments) was performed. The selected process

parameters to be varied were the air volume, the coating solution feed rate and the

inlet air temperature. Batch size, spraying pressure (set at 1.5 bar), solution

temperature and concentration were kept constant. The values corresponding to

different levels of the investigated process parameters are presented in table 1. The

9 experiments and corresponding process parameters levels are reported in table 2.

On the basis of the results obtained from these 9 experiments, one set of processing

conditions was selected and kept constant for further investigations related to the

effect of formulation parameters on the product’s properties and experimental

variance was assessed for that particular set of processing parameters.

After experiments E0 to E8, investigation of pore former inclusion within the

coating membrane and different coating levels on release rates was conducted.

7

Loaded carriers were coated with 100g mixture of ethylcellulose and PVP K30 in the

ratio of 9:1 or 9.5:0.5 (w/w). In these samples, 25% of dibutyl sebacate wrt. total

polymer weight was added as plasticizer. The above experiments are labelled as 5%

PVP K30 and 10% PVP K30. When discussing these results, E8 is labelled as 0%

PVP K30.

In addition to batches E0 to E8 which were coated with 100 g of ethylcellulose,

loaded carriers were also coated with polymeric membranes containing 40, 60 and

80 g of ethylcellulose, plasticized with 25% of dibutyl sebacate wrt. total polymer

weight. The expected coating material to total weight ratio equals 10%, 14.3%,

18.2% and 21.7% respectively for 40, 60, 80 and 100 g of ethylcellulose and the

corresponding products are noted L10, L14, L18 and L21 respectively.

2.2.3. Drug content determination

5 grams of sample was added to 200 ml of denatured ethanol to dissolve all

materials. Samples were withdrawn and analyzed by UV Spectrophotometry in the

linear range of the calibration curve (see 2.2.5).

2.2.4. Drug release experiments

Drug release assays were conducted using an SR8 Plus Dissolution Test Station

equipped with paddles rotating at 100 RPM (Hanson Research, Chartsworth, CA,

USA). 5 g sample was added to the 1 Liter dissolution medium at 37°C.

Demineralized water or a NaCl solution were used to investigate drug release. The

latter medium was used to test the effect of the medium’s osmolality.

Aliquots were withdrawn at predetermined time intervals, filtered using Chromophil

O-45/15MS filters (Duren, Germany) and replaced with fresh solvent. The withdrawn

8

quantities of solute were taken into consideration in the calculation of the

concentration. Concentration determination was done using UV spectrophotometry.

2.2.5. UV-spectrophotometry for paracetamol analysis

UV spectrophotometric analysis was performed using a Genesys 10S UV-Vis

spectrophotometer (Thermo Scientific, Erembodegem, Belgium). The calibration

curve was found to be linear between 125 mg/l and 1 mg/l (R²= 0.9998). The

samples were measured at 287 nm and 290 nm for aqueous and denatured ethanolic

solutions, respectively.

2.2.6. Particle size distribution

Particle size distribution for each batch was determined in the dry state using

laser light scattering (Mastersizer 2000, Malvern, UK). The particle size is

represented by the equivalent volume diameter D, and results are reported as first,

fifth and ninth deciles from the particle size distribution for each batch.

2.2.7. Scanning electron microscopy

The morphology and surface characteristics of unloaded, loaded, and coated

carriers was investigated. Scanning electron microscopy (SEM) images were

recorded using a Phillips XL30 SEM-FEG (Eindhoven, The Netherlands) equipped

with a Schottky field-emission electron gun. A beam of 15 kV was used and detection

was performed using a conventional Everhart-Thornley secondary electron detector.

The samples were affixed onto an aluminum stub with a double-sided adhesive

carbon tape and then coated with platinum under vacuum using a sputtering device

(Balzers Union, Liechtenstein) before imaging.

9

2.2.8. Statistical analysis

The data was tested using Multivariate analysis of variance (MANOVA). The test

was performed for equal variances and curve fittings were conducted using Minitab

17 (Minitab Inc., PA, USA).

3. Results and discussion

3.1. Drug loading

The loading method consisting of fluidizing and spraying the particles in a Wurster

system is mechanistically similar to the incipient wetness method which consists of

adding a volume of solution that is equal to the total pore volume leading to solution

penetration into the capillaries (Mellaerts et al., 2008). In our case, each particle

going through the spraying zone received a certain volume of drug solution that

ideally gets absorbed by the carrier and subsequently the particle exits the spraying

cone. The carrier loses solvent due to the drying capacity of the system and

eventually reaches back the spraying zone for another loading/drying cycle. The

product obtained at the end consists of mesoporous particles loaded with drug and

residual solvent. We observed considerable attrition due to friction in the fluid bed

system. Hence, in order to eliminate particles of size <300 µm which can cause

clogging of the nozzle during the coating phase, we had to introduce a sieving step.

The quantitative and qualitative composition of the products obtained from the

fluid bed after loading, coating and sieving are given in table 3. The solvent content

was determined by drying at 65 °C in a ventilated oven until constant mass was

reached. Compared to the starting materials, at the end of the third step 34.6 g of

silica and 15.4 g of paracetamol was lost. Out of this, 63.35% of silica and 33% of

paracetamol was lost during sieving. The remaining fraction was lost during loading

and coating.

10

There are numerous possible reasons for the loss of the material upon processing.

The attrition of carrier particles generates fines that serves as nuclei for

agglomeration and drug layering (Guignon et al., 2002). These particles are ejected

out of the system through the top filter afterwards. Although loss of drug as a

consequence of poor process conditions leading to spray drying of the drug solution

rather than drug loading in the silica can also occur, it is unlikely since spray drying

should typically be accompanied with the generation of a fine drug powder

(Broadhead et al., 1992), which was not observed during carrier loading in this study.

The theoretical drug loading by using a 4:1 carrier to drug weight ratio is 20%

w:w wrt. the final product weight. In our case, experimental drug loading obtained

was 18.8% w:w wrt. the total weight, which represents 94% of the value aimed for

initially.

Table 4 provides the particle size distribution from samples representative of all

prepared batches, including drug-free and drug-loaded silica particles. An overall

decrease in the particle size was observed upon drug loading which can be attributed

to both particle attrition due to mechanical stress upon fluidization, and to a powdery

fraction of the drug that remained on the surface of carrier particles after solvent

drying. This generates a smaller size fraction which can be observed on the surface

of drug loaded silica particles in figure 1, which overall appear to have a higher

rugosity as compared to the pure silica particles.

3.2. Drug content and effect of processing parameters on drug release

At the beginning of step 4, 450 g (dry weight) of the substrate was used for

coating. The dry weight of the product obtained at the end of phase 5 equals 525g

on average. The theoretical maximal dry weight equals 575g calculated by summing

up the substrate and the coating material dry weights, signifying that 50 g of material

11

was lost during steps 4 and 5. Paracetamol content for products E0 to E8 are

reported in table 5, and was found to equal 73.5g on average, corresponding to 14%

of the total weight.

Polymeric film coating of the loaded carrier particles was successful. This was

demonstrated by the gradual drug release in an aqueous medium from all the

batches as compared to the drug release from drug-loaded but uncoated silica

particles (figure 2). The cumulative drug release kinetics were different from one

batch to another, indicating that varying process parameters impacts end product

drug release characteristics. For instance, it was observed that increasing fluidizing

air inlet temperature from 35°C to 50°C resulted in a decrease in drug release, as

shown in figure 3, where high temperature batches release profiles are represented

in red, and low temperature batches in green. Decreasing coating solution feed rate

from 12 ml/min to 4 ml/min resulted in decreased drug release rates, as shown in

figure 4, where high feed rate batches release profiles are represented in red and low

feed rate batches release profiles in green. This can be explained by solvent

penetration of the loaded carrier upon spraying, which is likely to be higher when the

feed rate is at a higher level and temperature at a lower level. Upon penetration, the

solvent dissolves the drug which diffuses through the liquid phase, from the pores

towards the exterior due to a gradient in concentration, hence causing premature

drug release. After numerous spraying and drying cycles it results in a coating layer

that contains more drug in high feed rate batches than in low feed rate batches and

this results in a quicker release of the drug upon contact with water. Also, in this case

a more permeable membrane is expected since the drug present within the

membrane will act as a pore former. Obviously, the drug migration towards the

exterior of the carrier during coating is expected to be limited and decrease with time,

12

since the applied coating layer increases with time. This will “seal” the particle

thereby decreasing solvent penetration towards the core and drug diffusion towards

the coating layer.

From table 4, an increase in the overall particle size distribution upon coating can

be observed. This is attributed to the fact that upon spraying the coating solution,

small dusty particles are embedded within the coating layer resulting in a positive

shift of the overall particle size distribution. As expected, the higher coating solution

feed rate (12 ml/min) for batches E1, E2, E5 and E6 resulted in an overall higher

coated particle diameter as compared to the lower feed rate (4 ml/min) for batches

E3, E4, E7 and E8. The results clearly indicate that no lumps or aggregated coated

particles were formed and that individual silica particles were coated.

The factorial design’s surface responses are provided in figure 5, where drug

release at 2 hours was considered as a response, noted Y, and fluidizing air volume,

inlet air temperature and coating solution feed rate are noted X1, X2 and X3,

respectively. For the same reasons cited previously, response surfaces show an

increased drug release after 2 hours when feed rate is increased or inlet air

temperature is decreased. In comparison, fluidizing air volume has a lesser impact

on the drug release and is maximal when feed rate is at its highest level, with

increasing fluidizing air volume resulting in decreased drug release after two hours.

Figure 6 provides SEM images of samples from batches E8 and E2, produced at

low and high feed rate levels respectively. It is noticeable that batch E8 particles are

homogeneously coated with an apparently smooth polymeric layer and show little to

no surface defects. Batch E2 particles on the other hand appear to be rougher and

show a superior extent of defects, presumably due to solvent penetration into the

pores during coating and subsequent drug migration towards the coating layer, as

13

well as aggregation and subsequent breakage of aggregates which results in surface

cracks and therefore in a fast drug release.

From the 9 initial experiments, processing parameters corresponding to E8 were

chosen to be kept constant for further investigations. Experimental variance was

investigated for the above chosen set of processing conditions by producing a

replicate of E8 (labelled as E8D). Although apparently close in shape, the two drug

release curves were compared by means of multivariate analysis of variance (Figure

7). MANOVA was conducted using 4 time points at a time and then the two

remaining time points. The p-values for 1-4, 5-8 and 9-10 hours were 0.065, 0.749

and 0.141 respectively. Hence, no statistically significant difference was found for

time-points h1-h4, h5-h8 and h9-h10 separately. E8 and E8D release curves were

also compared by means of difference factor f1 and similarity factor f2, which

indicated similar drug release kinetics between the two batches (f1 = 4, f2 = 79).

Therefore the two release curves are considered to be statistically similar and the

experimental variance can be considered to be negligible for the experimental

conditions corresponding to experiment E8. This implies that for further experiments

where the effects of formulation variables were investigated, differences in release

profiles between batches was not caused by random fluctuations during the

processing stage and can therefore be attributed to variations in formulation

parameters.

3.3. Effect of pore former level on drug release

PVP K30 which is a water soluble polymer was incorporated in the coating layer

as a pore former. Upon contact with water, PVP K30 dissolves which results in

increased membrane permeability due to increased extent of drug release through

water filled pores. Figure 8 shows the drug release profiles from loaded mesoporous

14

particles coated with membranes containing 0% (E8), 5% and 10% PVP K30. The

drug content is reported in table 6. As expected drug release rate increases with

increased pore former level, indicating an increase of coating permeability due the

leaching out of PVP K30 from the coating layer which is similar to what is observed in

classical reservoir systems (Lecomte et al., 2003).

3.4. Effect of coating level on drug release

Drug release assays were conducted on L10, L14, L18 and E8 (labeled as L21 in

this figure), corresponding to coating levels 10%, 14%, 18% and 21% respectively.

Drug content and release for the batches L10, L14 and L18 are reported in table 6

and figure 9 respectively. Prolonged drug release from the reservoir systems is

typically comprised of a linear part during which part of the drug contained within the

coated cores is still undissolved resulting in a constant concentration gradient

between the inside and the outside of the reservoir and therefore a constant drug

release rate (Siepmann and Siepmann, 2012). During that phase, drug release is

given by the equation: M=K .D .S .C s

Lt

where K is the partition coefficient of the drug between the membrane and the

release medium, S is the total surface area of the rate controlling membrane, D is the

diffusivity of the drug in the coating membrane, CS is the saturation concentration of

the drug inside the reservoir, and L the thickness of the coating. The linear part of the

drug release profile is followed by a nonlinear part during which all of the drug

contained within the cores is dissolved and drug release rate decreases and reaches

zero eventually. A regression equation of the form y = ax + b was fitted to the drug

release data points corresponding to the linear part (Figure 10). Here, the slope ‘a’ is

the average drug release rate during the corresponding period of time, and the

15

deviation from the intercept ‘b’ corresponds to the amount of drug released

immediately upon contact with water, also called burst release. For batches L21, L18

and L14, zero order kinetics were maintained up to approximately 70% cumulative

drug release, corresponding to 6 hours, 5 hours and 2 hours durations of drug

release with R² values of 0.997, 0.998 and 0.989 respectively. On the other hand,

drug release up to 70% of drug content from L10 was linear during 40 minutes. It

was also observed that the computed average drug release rate for the considered

periods of time increased with higher coating levels, shifting from 11.8 %/h to 76.9

%/h when the coating level is reduced from 21 % to 10 %. The same trend is

observed for the computed burst release which increases from 2.4 % to 19.1% of

burst release when the coating level is reduced from 21 % to 10 %. The fact that the

drug release profile is linear and is maintained up to 70% of drug content for L21, L18

and L14 indicates that for these 3 coating levels, a virtually complete occlusion of

drug-containing mesopores by the coating was achieved (Muschert et al., 2009). This

results in the drug diffusion according to the previously described kinetics through the

pore-occluding polymeric membrane (Ozturk et al., 1990). On the other hand it

appears that a 10% coating level was not sufficient to efficiently occlude the majority

of the drug containing pores, as demonstrated by the poor linearity of the drug

release profile of the L10 batch and its relatively higher amount of drug burst release.

The SEM images shown in figure 11 and figure 5 (b, c) show noticeable

differences in terms of morphology and surface texture between the three coating

levels of 21%, 14% and 10%. An increase in smoothness and a decrease in apparent

defects with increasing coating levels is also evident which can be correlated with the

decreased drug release rate and burst release. This is expected since a defective

coating cannot sustain drug release thereby resulting in burst release effect. On the

16

other hand, decreasing surface roughness with increasing coating level indicates that

coating materials remain on the external surface and do not penetrate the mesopores

otherwise the external texture would have remained unchanged.

3.5. Effect of release medium’s osmolality on drug release

Investigation of the release medium osmolality’s effect on the release rate was

also conducted by preparing sodium chloride solutions of different molarities used as

release media. Experiments were conducted in duplicate. The concentrations of the

different solutions were 0.1, 0.2, 0.3 and 0.4 molar, corresponding to 0.2 osmol/kg,

0.4 osmol/kg, 0.6 osmol/kg and 0.8 osmol/kg. As it is expected from reservoir

systems (Ozturk et al., 1990), drug release kinetics were affected by the release

medium’s osmolality (figure 12). This is attributed to the osmotic pressure gradient

driven drug release mechanism. With increasing the release medium’s osmolality the

osmotic pressure gradient between the two boundaries of the rate controlling

membrane is decreased which in turn decreases the extent of drug release via

osmotic pumping.

4. Conclusion

In this study prolonged drug release was achieved based on commercially

available non-ordered and randomly (non-spherical) shaped mesoporous silica

particles as drug carriers. The carriers were successfully coated with an

ethylcellulose membrane from an organic solution using a pilot scale fluid bed system

as a unique processing unit. Based on the results of our experiments, drug contained

in the mesopores in such systems is isolated from the release medium with a

polymeric membrane that occludes the entrances of the pores which in turn aids in

modulating its release towards the external medium. Drug release from these loaded

and coated mesoporous materials followed the same trends usually observed with

17

classical reservoir systems. The unique formulation manufacturing approach we

investigated is potentially interesting for the preparation of modified/controlled

release from multiparticulate dosage forms for poorly soluble drugs. It requires only

one processing unit for the production of the coated multiparticulates. The

investigation of controlled drug release following the same principle can be further

continued for achieving site-specific or timed drug delivery of a single drug or a

combination of drugs.

Although in the present study, for process optimization, particles with a diameter

smaller than 300 µm were eliminated by sifting prior to coating, processing such

small particles would be feasible at a condition their size distribution is sufficiently

narrow.

One important remark is the significant attrition of the silica particles during

processing, which is a point of concern that needs to be tackled before this material

can be considered as reservoir carrier in fluid bed based manufacturing processes.

Acknowledgement

The authors would like to thank Mr. Duong Nhat Nguyen for his assistance with SEM.

Abhishek Singh acknowledges the financial support via an OT grant number

(OT/12/077) from KU Leuven.

18

References

Broadhead, J., Rouan, S.K.E., Rhodes, C.T., 1992. The spray drying of pharmaceuticals.

Drug Dev. Ind. Pharm. 18, 1169–1206. doi:10.3109/03639049209046327

Conley, R., Gupta, S.K., Sathyan, G., 2006. Clinical spectrum of the osmotic-controlled

release oral delivery system (OROS), an advanced oral delivery form. Curr. Med.

Res. Opin. 22, 1879–1892. doi:10.1185/030079906X132613

Felton, L.A., Porter, S.C., 2013. An update on pharmaceutical film coating for drug delivery.

Expert Opin. Drug Deliv. 10, 421–435. doi:10.1517/17425247.2013.763792

Guignon, B., Duquenoy, A., Dumoulin, E.D., 2002. Fluid Bed Encapsulation of Particles:

Principles and Practice. Dry. Technol. 20, 419–447. doi:10.1081/DRT-120002550

Lecomte, F., Siepmann, J., Walther, M., MacRae, R.J., Bodmeier, R., 2003. Blends of enteric

and GIT-insoluble polymers used for film coating: physicochemical characterization

and drug release patterns. J. Controlled Release 89, 457–471. doi:10.1016/S0168-

3659(03)00155-X

Limnell, T., Santos, H.A., Mäkilä, E., Heikkilä, T., Salonen, J., Murzin, D.Y., Kumar, N.,

Laaksonen, T., Peltonen, L., Hirvonen, J., 2011. Drug delivery formulations of ordered

19

and nonordered mesoporous silica: comparison of three drug loading methods. J.

Pharm. Sci. 100, 3294–3306. doi:10.1002/jps.22577

Mellaerts, R., Jammaer, J.A.G., Van Speybroeck, M., Chen, H., Humbeeck, J.V., Augustijns,

P., Van den Mooter, G., Martens, J.A., 2008. Physical State of Poorly Water Soluble

Therapeutic Molecules Loaded into SBA-15 Ordered Mesoporous Silica Carriers: A

Case Study with Itraconazole and Ibuprofen. Langmuir 24, 8651–8659.

doi:10.1021/la801161g

Muschert, S., Siepmann, F., Leclercq, B., Carlin, B., Siepmann, J., 2009. Prediction of drug

release from ethylcellulose coated pellets. J. Controlled Release 135, 71–79.

doi:10.1016/j.jconrel.2008.12.003

Nagane, K., Kimura, S., Ukai, K., Ogawa, N., Yamamoto, H., 2014. Practical approach to

prepare solid dispersion drug product using spherical silicate. Int. J. Pharm. 475,

364–371. doi:10.1016/j.ijpharm.2014.08.016

Ozturk, A.G., Ozturk, S.S., Palsson, B.O., Wheatley, T.A., Dressman, J.B., 1990. Mechanism

of release from pellets coated with an ethylcellulose-based film. J. Controlled Release

14, 203–213. doi:10.1016/0168-3659(90)90160-U

Pérez-Esteve, É., Fuentes, A., Coll, C., Acosta, C., Bernardos, A., Amorós, P., Marcos, M.D.,

Sancenón, F., Martínez-Máñez, R., Barat, J.M., 2015. Modulation of folic acid

bioaccessibility by encapsulation in pH-responsive gated mesoporous silica particles.

Microporous Mesoporous Mater. 202, 124–132.

doi:10.1016/j.micromeso.2014.09.049

Qu, F., Zhu, G., Huang, S., Li, S., Sun, J., Zhang, D., Qiu, S., 2006. Controlled release of

Captopril by regulating the pore size and morphology of ordered mesoporous silica.

Microporous Mesoporous Mater. 92, 1–9. doi:10.1016/j.micromeso.2005.12.004

Salsa, T., Veiga, F., Pina, M.E., 1997. Oral Controlled-Release Dosage Forms. I. Cellulose

Ether Polymers in Hydrophilic Matrices. Drug Dev. Ind. Pharm. 23, 929–938.

doi:10.3109/03639049709148697

20

Siepmann, J., Siepmann, F., 2012. Modeling of diffusion controlled drug delivery. J.

Controlled Release, Drug Delivery Research in Europe 161, 351–362.

doi:10.1016/j.jconrel.2011.10.006

Singh, A., Worku, Z.A., Van den Mooter, G., 2011. Oral formulation strategies to improve

solubility of poorly water-soluble drugs. Expert Opin Drug Deliv 8, 1361-1378.

Song, S.-W., Hidajat, K., Kawi, S., 2005. Functionalized SBA-15 Materials as Carriers for

Controlled Drug Delivery:  Influence of Surface Properties on Matrix−Drug

Interactions. Langmuir 21, 9568–9575. doi:10.1021/la051167e

Tang, E.S.K., Chan, L.W., Heng, D.P.W.S., 2012. Coating of multiparticulates for sustained

release. Am. J. Drug Deliv. 3, 17–28. doi:10.2165/00137696-200503010-00003

Tiwari, S., Rajabi-Siahboomi, A., 2008. Extended-Release Oral Drug Delivery Technologies:

Monolithic Matrix Systems, in: Jain, K. (Ed.), Drug Delivery Systems, Methods in

Molecular BiologyTM. Humana Press, pp. 217–243.

Vallet-Regi, M., Rámila, A., del Real, R.P., Pérez-Pariente, J., 2001. A New Property of

MCM-41:  Drug Delivery System. Chem. Mater. 13, 308–311.

doi:10.1021/cm0011559

Vialpando, M., Aerts, A., Persoons, J., Martens, J., Van den Mooter, G., 2011. Evaluation of

ordered mesoporous silica as a carrier for poorly soluble drugs: Influence of pressure

on the structure and drug release. J. Pharm. Sci. 100, 3411–3420.

doi:10.1002/jps.22535

21

Captions to figures

Figure 1: SEM images of (a) pure silica particles and (b) drug-loaded silica particles.

Figure 2: Drug release profiles for batches E0 to E8. Error bars indicate standard

deviations (n=3).

Figure 3: Drug release profiles for batches E1 to E8. Curves with same symbols differ

only in feed rate with red and green corresponding to high and low feed rate levels

respectively. Error bars correspond to standard deviations (n=3).

Figure 4: Drug release profiles for batches E1, E2, E5 and E6, curves with same

symbols differ only in temperature with red and green corresponding to high and low

temperature levels respectively. Error bars indicate standard deviations (n=3).

Figure 5: Full factorial design’s response surfaces, with Y = drug release at 2 h, X1 =

Fluidizing air volume, X2 = Inlet air temperature, X3 = Feed rate.

Figure 6: SEM images of particles from batches: (a) E2 and (b) and (c) E8.

Figure 7: Drug release profiles for batches E8 and E8D. Error bars indicate standard

deviations (n=3).

Figure 8: Drug release profiles from loaded particles coated with 0%, 5% and 10%

PVP K30. Error bars indicate the standard deviation (n=3).

Figure 9: Drug release profiles from samples with different coating levels. Error bars

indicate standard deviations (n=3).

Figure 10: Linear parts of drug release profiles from batches L10, L14, 18 and L21

and corresponding fitting factors.

Figure 11: SEM images of particles from batches L10 (a, b, c) and L14 (d, e, f).

22

Figure 12: Drug release profiles from E8 in media with different osmolalities. Error

bars indicate the data range (n=2).

23

Fig1

24

Fig2

25

Fig 3

26

Fig 4

27

Fig 528

Fig 6

29

Fig 7

30

Fig 8

31

Fig 9

32

Fig 10

33

Fig 11

34

Fig 12

35

Tables

Table 1: Variables and levels of the full factorial design.

Variable level Air volume (Pa) Temperature (°C) Feed rate

(ml/min)

+ 50.0 50.0 12.0

0 37.5 42.5 8.0

- 25.0 35.0 4.0

36

Table 2: Experimental design runs and corresponding levels.

Experiment Air volume Temperature Feed rate

E0 0 0 0

E1 + + +

E2 - - +

E3 - + -

E4 + - -

E5 - + +

E6 + - +

E7 + + -

E8 - - -

37

Table 3: Composition of products at the end of steps 2 and 3 (Value ± standard deviations, n= 3).

Time Carrier (g) Solvent (g) Paracetamol (g)

Time zero 400.0 0.0 0.0

End of phase 2 387.3 ± 7.9 53.0 ± 1.0 89.7 ± 1.5

End of phase 3 365.4 ± 7.7 50.0 ± 1.0 84.6 ± 1.4

38

Table 4: First, fifth and ninth deciles from particle size distributions of various

batches (Value ± standard deviations, n= 3).

Batch D(0,1)

(µm)

D(0,5)

(µm)

D(0,9)

(µm)

Batch D(0,1)

(µm)

D(0,5)

(µm)

D(0,9)

(µm)

Pure

silica

178 ± 7 371 ± 8 630 ± 7 E7 356 ± 4 484 ± 7 655 ± 9

Uncoat

ed

109 ± 4 314 ± 7 569 ±

10

E8 334 ± 6 454 ± 8 616 ± 7

E0 370 ±

10

509 ± 8 704 ±

11

E8D 342 ± 8 466 ± 5 634 ± 8

E1 355 ± 8 485 ± 9 661 ± 8 L18 331 ± 6 453 ± 7 619 ± 9

E2 389 ± 6 544 ±

10

775 ± 9 L14 326 ± 5 443 ± 8 602 ± 7

E3 328 ± 4 445 ± 6 603 ± 6 L10 327 ± 8 450 ± 6 624 ±

10

E4 352 ± 7 481 ± 8 653 ± 7 PVP5 339 ± 7 465 ± 7 638 ± 9

E5 350 ± 6 480 ±

11

656 ± 8 PVP10 323 ± 5 441 ± 6 597 ± 7

E6 362 ±

10

497 ±

10

680 ± 9

39

Table 5: Drug content in batches E1 to E8D.

Batch E1 E2 E3 E4 E5 Averag

e

Total drug content (g) 76.6 73.3 71.9 71.4 76.2 73.5

Total:drug weight

ratio (%)

14.6 14.0 13.7 13.6 14.5 14.0

Batch E6 E7 E8 E0 E8D SD

Total drug content (g) 74.8 73.7 73.1 72.4 71.3 1.9

Total:drug weight

ratio (%)

14.2 14.0 13.9 13.8 13.6 0.4

40

Table 6: Drug contents in batches 5% PVP to L18.

Batch 5% PVP 10%PVP L10 L14 L18

Total drug content (g) 73.8 72.8 71.8 72.7 76.6

Total:drug weight ratio

(%)

14.0 13.9 16.0 15.3 15.3

41