Supercritical fluid-assisted preparation of imprinted contact lenses for drug delivery

Acta Biomaterialia 7 (2011) 1019–1030

Contents lists available at ScienceDirect

Acta Biomaterialia

journal homepage: www.elsevier .com/locate /actabiomat

Supercritical fluid-assisted preparation of imprinted contact lenses for drug delivery

Fernando Yañez a, Lahja Martikainen b, Mara E.M. Braga b, Carmen Alvarez-Lorenzo a,⇑, Angel Concheiro a,Catarina M.M. Duarte c,d, Maria H. Gil b, Hermínio C. de Sousa b,⇑a Departamento de Farmacia y Tecnología Farmacéutica, Facultad de Farmacia, Universidad de Santiago de Compostela, 15782-Santiago de Compostela, Spainb CIEPQPF, Chemical Engineering Department, FCTUC, University of Coimbra, Rua Sílvio Lima, Pólo II–Pinhal de Marrocos, 3030-790 Coimbra, Portugalc Nutraceuticals and Delivery Laboratory, Instituto de Biologia Experimental e Tecnológica, Apartado 12, 2781-901 Oeiras, Portugald Instituto de Tecnologia Química e Biológica, Universidade Nova de Lisboa, Avenida da Republica, 2780-157 Oeiras, Portugal

a r t i c l e i n f o

Article history:Received 30 July 2010Received in revised form 22 September 2010Accepted 4 October 2010Available online 8 October 2010

Keywords:Therapeutic contact lensesMolecular imprintingSupercritical fluid technologyFlurbiprofenControlled drug release

1742-7061/$ - see front matter � 2010 Acta Materialdoi:10.1016/j.actbio.2010.10.003

⇑ Corresponding authors. Tel.: +351 239 798 749; fE-mail addresses: [email protected] (

eq.uc.pt (H.C. de Sousa).

a b s t r a c t

The aim of this work was to develop an innovative supercritical fluid (SCF)-assisted molecular imprintingmethod to endow commercial soft contact lenses (SCLs) with the ability to load specific drugs and to con-trol their release. This approach seeks to overcome the limitation of the common loading of preformedSCLs by immersion in concentrated drug solutions (only valid for highly water soluble drugs) and ofthe molecular imprinting methods that require choice of the drug before polymerization and thus to cre-ate drug-tailored networks. In particular, we focused on improving the flurbiprofen load/release capacityof daily wear Hilafilcon B commercial SCLs by the use of sequential SCF flurbiprofen impregnation andextraction steps. Supercritical carbon dioxide (scCO2) impregnation assays were performed at 12.0 MPaand 40 �C, while scCO2 extractions were performed at 20.0 MPa and 40 �C. Conventional flurbiprofensorption and drug removal experiments in aqueous solutions were carried out for comparison purposes.SCF-processed SCLs showed a recognition ability and a higher affinity for flurbiprofen in aqueous solutionthan for the structurally related ibuprofen and dexamethasone, which suggests the creation of molecu-larly imprinted cavities driven by both physical (swelling/plasticization) and chemical (carbonyl groupsin the network with the C–F group in the drug) interactions. Processing with scCO2 did not alter some ofthe critical functional properties of SCLs (glass transition temperature, transmittance, oxygen permeabil-ity, contact angle), enabled the control of drug loaded/released amounts (by the application of severalconsecutive processing cycles) and permitted the preparation of hydrophobic drug-based therapeuticSCLs in much shorter process times than those using conventional aqueous-based molecular imprintingmethods.

� 2010 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.

1. Introduction other polymeric matrices, SCLs can be worn more frequently and

Conventional ophthalmic drug delivery systems (DDS), such astopical eye drops, usually lead to poor ocular drug bioavailabilitydue to their short residence time on the cornea surface and the effi-cient defense mechanisms of the eye [1]. Drug formulation in nano-particle/nanocapsule suspensions, liposomes, collagen shields orocular inserts may enable the achievement of greater drug concen-trations in the ocular tissues [2,3]. The potential of soft contactlenses (SCLs) as ophthalmic DDS was first tested by Sedlacek [4],and later by Waltman and Kaufman [5] and by Jain [6]. SCLs loadedby immersion in a drug solution (or suspension/emulsion) providemore sustained drug levels in tears and also in the post-lens tearfilm fluid, compared with topical drop instillation, thus enhancingdrug cornea permeation and minimizing the occurrence of unto-ward systemic absorption [7–10]. Moreover, and compared with

ia Inc. Published by Elsevier Ltd. A

ax: +351 239 798 703.C. Alvarez-Lorenzo), hsousa@

for longer periods because of their excellent biocompatibility, com-fort and patient compliance/acceptance. Medicated SCLs may com-bine the role of DDS with the correction of refractive deficiencies,acting as a combination product according to FDA regulations[11]. Nevertheless, few drugs can be effectively loaded into com-mercial SCLs, which also lack control of the drug delivery onceplaced on the cornea [12–15]. To overcome these problems, and alsoto attain more sustained drug release profiles, several approacheshave been proposed [9,15–17], among which the use of molecularlyimprinted SCLs has been suggested as an improved biomimeticmethod for the preparation of therapeutic SCLs [8,15,18–20].

The traditional molecular imprinting method consists of addinga template molecule (e.g. a drug) to a monomeric solution in orderto induce spatial arrangement of the monomers according to theirinteraction capabilities with the template [21–25]. Subsequentpolymerization and cross-linking fix such spatial assemblies and,after removal of the template, the resultant polymeric networkexhibits ‘‘cavities” with sizes and shapes specific for that template.Once again in contact with the template the polymer networks can

ll rights reserved.

1020 F. Yañez et al. / Acta Biomaterialia 7 (2011) 1019–1030

easily re-uptake it into the previously formed specific cavities. Inaddition, release of the template is performed in a more sustainedmanner, owing to the strength of the polymer–drug interactions[15,18,20,22,24]. Two important concerns in the preparation ofmolecularly imprinted polymers (MIPs) for DDS are that the tem-plate (i.e. the drug of interest) has to be stable under the polymer-ization conditions and that no toxic solvents should remain on thedevice. In the particular case of SCLs, the cross-linking density hasto be compatible with the required flexibility of the material andthus the imprinted cavities should possess high affinity for the drugin order to compensate for their lower physical stability [24,26,27].

Despite traditional molecular imprinting methods involving theformation of a polymer network in the presence of the template, ifwe consider only the phenomena involved (polymer–templateinteraction) and the final result obtained (‘‘cavities” of adequateshape/volume, chemical affinity and recognition ability) the term‘‘imprinting” may be used at any time to indicate a specific ‘‘im-print” is produced in a certain structure by a specific template mol-ecule at the molecular (nano)scale [25,28,29]. Structurallycommercial SCLs comprise several co-monomers and cross-linkers,each of which has specific chemical and physical functionalities.Due to the low cross-linking density, there remains available chainmobility and free volume between the polymeric chains (other-wise, SCLs would not swell upon water sorption). Thus, these loosechains, together with any other side-chain branches, can still be‘‘reorganized” and even ‘‘fixed” when a template molecule is incor-porated into the network and can establish specific interactionswith certain regions of the polymer. These physical rearrange-ments, organization and imprinting phenomena, also known aspost-imprinting, have been demonstrated in other low density,cross-linked hydrogels [22,29–32].

It is well known that some drugs may be deposited in (orextracted from) polymeric matrices by dissolving them in com-pressed highly volatile fluids (like carbon dioxide) at temperaturesand pressures near or above their critical values and placing theresulting mixtures in contact with those polymeric matrices. Underthese conditions the compressed fluid mixture can act also as aplasticizing agent, causing swelling of the polymeric network andhelping drug diffusion into (or out of) it [33–40]. Supercritical car-bon dioxide (scCO2) is the most commonly used SCF because it isan abundant and cheap, non-flammable, relatively inert and ‘‘gen-erally recognized as safe” (GRAS) solvent, soluble in aqueous mediaand it can plasticize and decrease the glass transition temperatureof most polymeric materials. Furthermore, it has a low critical tem-perature (31.05 �C), allowing working at relatively low tempera-tures, suitable for thermally labile substances [33–35]. Comparedwith common liquid plasticizers, scCO2 penetrates deeper intodense polymeric networks and can be easily removed after pro-cessing [36,41]. Supercritical solvent impregnation (SSI) and super-critical fluid extraction (SFE) using scCO2 have already shownadvantages for the development of drug impregnated polymericmaterials which can be used as DDSs for many biomedical applica-tions [42,43]. SSI allows the homogeneous impregnation of poly-meric matrices with drugs in a relatively short time and, whenproperly employed, it does not alter and/or damage their physical,chemical, and mechanical properties. Drug loading and even depthof penetration can be easily controlled by the regulation of severaloperational conditions [44–48]. Moreover, scCO2 can also beemployed to extract undesired residual solvents, oligomers ormonomers, or any other additives present in polymeric materials[34–39,45,49].

In the present work we explore the suitability of applying SCF-based technologies, namely SSI and SFE, in an innovative methodto induce specific nano-range structural changes in preformedcommercially available SCLs with the aim of creating specific andhigh affinity ‘‘cavities” for a template drug. By doing this, we ex-

pect to enhance drug loading and to achieve more sustained/ex-tended delivery. SCFs have already been used in the preparationof MIPs, but only as the solvent for emulsion polymerization ofthe co-monomer mixture [50]. To the best of our knowledge thisis the first time the technique has been used to induce a post-imprinting modification. It is expected that scCO2 plasticizes andswells wet SCLs and facilitates the template impregnation/extrac-tion processes, as well as potential rearrangement of the polymericchains. In addition, dissolution of scCO2 inside water swollen SCLsmay decrease the internal pH (due to the formation of carbonicacid) and the total polarity of water inside SCLs [47,48]. Dependingon the polymeric material involved, this may also affect the poly-mer ionization state and the chain conformation, as well as the sol-ubility of the drugs to be impregnated/extracted, especially if theyhave low polarity or their solubility depends on pH. Under slowdepressurization the CO2 is vented, leaving the drug molecularlydispersed inside the SCL polymeric matrix. The pH is also expectedto return to its initial value. Drug removal, by SCF extraction orimmersion in aqueous medium, and subsequent re-uptake experi-ments may provide evidence of the previous formation of cavitieswith a size and shape complementary to those of the templatedrug molecules. If the conformation of the cavities created is stablethe ability of the lens to load the template drug should be greaterthan before processing and such an improvement should be main-tained or even increased after successive sequential reloading andrelease steps. On the other hand, and also due to the specific cav-ities created, drug release should be more sustained. Thus wemay consider that the preformed polymeric networks were ‘‘drugimprinted” on the molecular scale by the use of this innovativeSCF-assisted molecular imprinting method. The experiments werecarried out with daily wear disposable Hilafilcon B commercialcontact lenses (Bausch & Lomb�) using the poorly water solublenon-steroidal anti-inflammatory (NSAID) flurbiprofen as template(Fig. 1). One driving force of post-imprinting may be the interac-tion between the carbonyl groups of these SCLs and the C–F groupof the drug, since this type of interaction is known to play a keyrole in many biorecognition processes in vivo [51,52]. In additionto flurbiprofen loading and release experiments, the competitivebinding of structurally similar drugs, ibuprofen and dexametha-sone, was analyzed in order to determine whether the SCF pro-cesses employed were able to produce SCLs with improved andspecific affinity for flurbiprofen.

2. Materials and methods

Hilafilcon B SCLs (Soflens, 59 wt.% water content, 8.6 mm basecurve, �8.00D power, 14.2 mm Ø) were kindly supplied by Bausch& Lomb� (Lisbon, Portugal) and by Mart-Optic (Coimbra, Portugal).Flurbiprofen (P99%, TLC), ibuprofen (P98%, HPLC) and dexameth-asone (P98%, HPLC) were from Sigma–Aldrich (Barcelona, Spain).Carbon dioxide (99.998%) was from Praxair (Spain) and ethanol(P99.0%) from Panreac Química, (Barcelona, Spain). Purified water(MilliQ�, Millipore, resistivity >18 MO cm) was obtained by reverseosmosis. Before use the Hilafilcon B SCLs were thoroughly washedwith Milli-Q water (under stirring for 12 h) in order to remove ab-sorbed substances from the SCL liquid storage solution. WashedSCLs were then stored at room temperature in a controlled �83%relative humidity environment (saturated potassium chlorideatmosphere). The same storage procedure was employed for allprocessed SCLs in this work.

2.1. Flurbiprofen loading/extraction procedures

Commercial SCLs were loaded with flurbiprofen by twoapproaches: (i) supercritical solvent impregnation (SSI) or (ii)

Fig. 1. Chemical structures of (A) the drugs employed and (B) the constitutive monomers, co-monomers and cross-linkers of Hilafilcon B contact lenses. HEMA, 2-hydroxyethyl methacrylate; MAA, methacrylic acid; EGDMA, ethyleneglycol dimethacrylate; NVP, N-vinylpyrrolidone.

F. Yañez et al. / Acta Biomaterialia 7 (2011) 1019–1030 1021

immersion in aqueous drug solutions. Drug removal was per-formed using supercritical fluid extraction (SFE) or leaching/extraction in water. Combined sequential loading/removal proce-dures (three consecutive cycles) were employed: SSI followed bySFE or water immersion followed by water leaching (conventionalmethod). The loading/removal processing cycles are shown in theflowchart depicted in Fig. 2.

2.1.1. Conventional drug loading by immersion in drug aqueoussolution

SCLs were placed in Falcon tubes containing 40 ml of flurbipro-fen aqueous solution (8 mg l�1) and stirred (100 rpm and 37 �C) for

Fig. 2. Schematic representation of the SCLs d

2.5 h. Drug loading kinetics experiments were also carried out un-der similar conditions, monitoring the decrease in the concentra-tion of flurbiprofen in the solution at various time points up to13 h. All experiments were performed in triplicate.

2.1.2. Drug leaching/extraction and drug release experimentsSCLs loaded by conventional immersion in an aqueous drug

solution were placed in 80 ml of purified water and the drug con-centration was then monitored spectrophotometrically at247.5 nm (model V-650, Jasco, Japan) for 8 h. The medium was re-placed by fresh purified water at predetermined time intervals(every 24 h). Leaching was carried out until no flurbiprofen could

rug loading/removal methods employed.

1022 F. Yañez et al. / Acta Biomaterialia 7 (2011) 1019–1030

be detected (4 days). For the release experiments SCLs previouslyloaded with flurbiprofen, either by SSI or by immersion in aqueousdrug solutions, were immersed in 80 ml of purified water for 8 h. Inthis case, 2.5 ml aliquots were collected at predetermined timeperiods (without replenishment of the aqueous medium) and thereleased drug was quantified using the same spectrophotometricmethod as described above. All assays were carried out in tripli-cate, at 37 �C and under stirring (100 rpm).

2.1.3. Supercritical solvent impregnation (SSI) drug loading methodSSI experiments were performed at 40 �C and 12 MPa using a

previously described discontinuous supercritical impregnationapparatus [42–48]. CO2 was introduced into and pressurized in atemperature controlled high pressure impregnation cell loadedwith 3.5 mg of flurbiprofen. Six commercial Hilafilcon B SCLs wereimpregnated per experiment. Contact lenses were placed insidethe high pressure stainless steel cell (internal volume10 � 10�6 m3), fitted in a stainless steel support in order to main-tain two SCLs at three different heights inside the cell. The amountof flurbiprofen was �5.5 times greater than the amount needed tosaturate scCO2 under the operational process conditions [53]. Mag-netic stirring (700 rpm) was used to solubilize the drug andhomogenize the high pressure mixture. After a 2.5 h impregnationperiod the compressed fluid was removed by slow expansion in or-der to prevent the drug-loaded SCLs from being damaged. Theaverage depressurization rate was 0.06 MPa min�1. Drug-loadedSCLs were then recovered in a semi-dry state and stored until fur-ther processing/analysis.

SCLs loaded with flurbiprofen by SSI underwent two differentdrug removal procedures: (a) extraction with scCO2 (by SFE) andthen washing in water (to quantify the residual flurbiprofenremaining in the lenses); (b) direct immersion in water in orderto record the drug release profiles. SCLs subjected to drug removalprocedure (a) were re-impregnated with flurbiprofen using thesame SSI procedure described above and then again subjected tothe SFE extraction and release in water tests (see Fig. 2). In total,three successive SSI loading and SFE/water removal cycles werecarried out. All experiments were carried out in triplicate.

2.1.4. Supercritical fluid extraction (SFE) drug removal methodFlurbiprofen extraction/removal experiments from drug-loaded

SCLs (by SSI) were carried out using a previously described SFEapparatus [45]. Drug-loaded SCLs were fitted in a stainless steelsupport and placed inside a high pressure stainless steel extractioncell (internal volume 30 � 10�6 m3), which was immersed in atemperature controlled water bath at 40 �C. scCO2 was introducedand the pressure maintained for 14 h at 12 MPa (static swellingand extraction period). Then the pressure was increased to20.0 MPa and continuous scCO2 extraction was carried out for 5 hat a constant scCO2 flow rate (�0.1 l min�1). Over the continuousextraction period, and considering the extraction duration, thescCO2 flow rate and flurbiprofen drug solubility in scCO2 underthe operational conditions, the total amount of scCO2 that passedthrough the cell was calculated to be almost 200 times that neces-sary to solubilize the introduced flurbiprofen. The system was thenslowly depressurized (�0.1 l min�1) and the outlet CO2 effluentwas passed/bubbled through 10 ml of ethanol retained in a coldtrap. Extracted SCLs were then recovered and, at the end of exper-iment, the equipment tubing lines were washed with 200 ml ofethanol. All flurbiprofen dissolved in ethanol was later concen-trated in a Multivapor™ under vacuum (Büchi, Flawil, Switzerland)and quantified spectrophotometrically at 247.5 nm (model V-650,Jasco, Japan). Experiments were carried out for three contact lensesper experiment.

Any remaining flurbiprofen (termed residual drug) still presentin the SFE-extracted SCLs was later removed by immersion in

80 ml of water (at 37 �C and 100 rpm), replacing the medium every24 h and monitoring the drug concentration spectrophotometri-cally at 247.5 nm (model V-650, Jasco, Japan). The experimentswere carried out in triplicate until no flurbiprofen could be de-tected in the water (4 days).

2.2. Molecular imprinting proof of concept

SCLs processed for three SSI/SFE cycles and unprocessed SCLs(as controls) were immersed in 20 ml of an aqueous solution of3.3 � 10�5 M flurbiprofen, ibuprofen or dexamethasone for 14 h.Drug concentration in the medium was spectrophotometricallymonitored during the loading process and in the subsequent re-lease experiments in water. All assays were carried out induplicate.

2.3. SCLs characterization

2.3.1. Surface morphologyDried, non-processed and processed SCLs were coated with gold

(approximately 300 ÅA0

), in an argon atmosphere and then observedby scanning electron microscopy (SEM) (model JSM-5310, Jeol, Ja-pan) at 20 keV, at various magnifications.

2.3.2. Fourier transform infrared (FTIR) attenuated total reflection(ATR) spectroscopy

Non-processed and scCO2 processed SCLs (three cycles of SSI/SFE) were analysed in the 400–4000 cm�1 range using a GladiATRsingle reflection spectrometer (Pike Technologies, Madison, WI) for32 scans at 2 cm�1 resolution.

2.3.3. Differential scanning calorimetry (DSC)Experiments were carried out in duplicate using a DSC Q100 (TA

Instruments, New Castle DE) with a refrigerated cooling accessory.Nitrogen was used as the purge gas at a flow rate of 50 ml min�1.The calorimeter was calibrated for baseline using no pan, for cellconstant and temperature using indium (melting point 156.61 �C,enthalpy of fusion 28.71 J g�1), and for heat capacity using sapphirestandards. To determine the glass transition temperature Tg exper-iments were performed using non-hermetic aluminium pans, inwhich 5–10 mg dried disks were accurately weighed, then coveredwith a lid and program heated from 30 to 100 �C, cooled to 0 �C andfinally heated again up to 300 �C, at 10 �C min�1.

2.3.4. Water content and swelling capacityNon-processed and scCO2 processed SCLs were dried in a forced

air convection oven for 72 h at 40 �C. Then the SCLs were immersedin water and their weights measured every 24 h until a constantweight was achieved (generally after 5 days). The experimentswere carried out in triplicate.

2.3.5. Oxygen permeabilityNon-processed and scCO2 processed SCLs were swollen in 0.9%

NaCl solution and their oxygen permeabilities measured in tripli-cate using a Createch permeometer (model 210T, Rehder Develop-ment Co., Castro Valley, CA) at room temperature and 100% relativehumidity.

2.3.6. Contact angle/surface free energy measurementsThe contact angles of Milli-Q water, ethylene glycol and form-

amide on non-processed and scCO2 processed SCLs were evaluatedin four regions/quadrants of each lens (previously cut into fourpieces). Measurements were performed using the sessile drop (6–7 ll) method and an OCA20 contact angle apparatus (DataphysicsInstruments GmbH, Germany). The surface energy was estimated

F. Yañez et al. / Acta Biomaterialia 7 (2011) 1019–1030 1023

using the extended Fowkes (xF) method, which is generally recom-mended for modified polymer surfaces [54,55].

2.4. Model fitting and statistical analysis

2.4.1. Relative amounts of released to loaded flurbiprofenThe percentage of drug released was calculated according to the

equation:

Relative released=loaded drug ð%Þ

¼ Released drugðLoaded drugþ Residual drugÞ � 100 ð1Þ

where ‘‘released drug” is the amount of flurbiprofen (per mg wetlens) released in water during an 8 h release period, ‘‘loaded drug”is the amount of flurbiprofen (per mg wet lens) extracted by waterleaching or by SCF extraction, and ‘‘residual drug” is the amount offlurbiprofen (per mg wet lens) still present in the SFE-extractedSCLs and that was removed by leaching in water.

2.4.2. Diffusion coefficientsDrug release profiles below 60% of the amount released were

fitted to Eq. (2) [56–58]:

Mt

M1¼ ktn ð2Þ

where Mt and M1 represent the cumulative (absolute) amount ofdrug released at time t and at infinite time, respectively, k is akinetic constant that incorporates the structural and geometriccharacteristics of the delivery device (polymer + drug), and n isthe release exponent, which can provide information about the drugrelease mechanism. The following equations were also used [58]:

Mt

M1¼ 4

Dt

pl2

� �12

ð3Þ

Mt

M1¼ 1� 8

p2

� �exp �p2Dt

l2

� �ð4Þ

where l is the thickness of the sample and D is the diffusion coeffi-cient, which is assumed to be constant. Eq. (3) is only valid for thefirst 60% of the total release (Mt/M1 60.6) while Eq. (4) is valid forthe last 40% of the total release (Mt/M1 P0.4) and has been widelyused to obtain diffusion coefficients from drug release experimentaldata. After linearization and regression analysis of these equationsthe drug diffusion coefficients D can be determined from the result-ing slopes.

2.4.3. Statistical analysisStatistical analyses were carried out using analysis of variance

(ANOVA) and the Tukey HSD test in Statistica 5.0 (StatSoft Inc.,Tulsa, OK).

3. Results and discussion

In the present work sequential steps comprising scCO2 impreg-nation (to load the drug template) and scCO2 extraction (to removethe drug template) were applied, and the effects of these succes-sive steps on the physical and mechanical properties of commer-cially available SCLs, as well as on their performance as drugdelivery devices, were evaluated. Flurbiprofen was chosen as thedrug template because this NSAID is widely used to treat commonocular inflammatory processes and after eye surgery. Its pooraqueous solubility (5 � 10�5 M or 12 mg l�1) [59–60] hinders itsuptake by SCLs from aqueous solution. However, scCO2 has beenshown to be a useful solvent for several hydrophobic drugs [61],including flurbiprofen [53]. Thus, the first aim of the work was to

elucidate how much flurbiprofen could be impregnated into dailywear Hilafilcon B commercial SCLs (Bausch & Lomb�) using scCO2

at 12.0 MPa and 40 �C, compared with conventional loading byimmersion in a saturated aqueous drug solution. The flurbipro-fen-impregnated SCLs were divided into two groups (as shown inFig. 2): one was used to check the drug release kinetics in water,while the other group underwent extraction with scCO2 followedby washing in water in order to quantify the total amount of drugthat had been loaded. Some lenses underwent successive cycles ofimpregnation/extraction with scCO2.

3.1. Flurbiprofen loading/impregnation and release/extraction

The total amount of flurbiprofen impregnated using scCO2 wasestimated as the sum of the amount extracted by scCO2 plus a min-or amount that was remained and could be removed with water(Fig. 3A, grey columns). It is evident that consecutive SSI/SFEprocessing cycles led to greater amounts of flurbiprofen beingimpregnated, i.e. the amount of drug in the lenses was increasedalmost 2-fold for two cycle processing over a single processingcycle and further increased 2-fold more after three cycle process-ing. From the one cycle to the three cycles procedure the loadingcapacity of Hilafilcon B SCLs increased approximately 450%, i.e. to55 mg g�1. In contrast, similar loading/extraction cycles carriedout in aqueous medium (i.e. immersion in a drug solution followedby release in water) led to a significantly lower amount loaded(<1 mg g�1) and consecutive loading/extraction steps did not im-prove the yield but decreased it, probably due to irreversible bind-ing of the hydrophobic drug to the network, which prevents moremolecules being adsorbed. If one compares the time required forthe processing of the lenses (drug loading/removal) with scCO2

(less than 1 day) and with water-based solutions (4–5 days), theadvantage of the former procedure is even more evident.

Commercial ophthalmic drops contain 0.03% flurbiprofen, i.e.each drop contains about 15 lg of flurbiprofen. Therefore, eachscCO2 processed SCL (43–44 mg weight) contained the sameamount of drug as 29 or 160 drops of the ophthalmic solution,after one or three processing cycles. SCLs loaded by immersionin water took up the equivalent of only 2.5 drops of flurbiprofensolution.

The overall efficiency of SCF-based processes is governed by theoperational conditions employed (pressure, temperature, process-ing time, co-solvent addition and composition, flow rates, depres-surization rate, etc.) and by the physico-chemical interactionsestablished between the substances involved (the lens, scCO2,water, and drug) in the process [36,38,41,45–48]. In the presentcase the most relevant interactions to be considered are thescCO2–flurbiprofen interactions (which determine flurbiprofen sol-ubility in scCO2), the water swollen SCL–scCO2 interactions (whichlead to lens swelling and plasticization), and the flurbiprofen–water swollen SCL interactions (which control flurbiprofen solubil-ity and partitioning in the lens) [36,38,41,47,48]. scCO2 is a non-polar solvent (despite the fact that, under certain circumstances,it can form quadrupoles) which has a higher affinity for low polar-ity small hydrophobic drugs (such as flurbiprofen) than water.Therefore, the scCO2–flurbiprofen interactions are favourable and,under the operational conditions employed, scCO2 dissolves moreflurbiprofen molecules per unit volume than water; drug solubilitywas estimated as �160 and �510 mg l�1 at 40 �C in scCO2 duringSSI (12 MPa) and SFE (20 MPa), respectively, and �12 mg l�1 at25 �C in water (0.1 MPa). This means that flurbiprofen can be rap-idly dissolved and loaded into a scCO2 phase, which is a helpful fea-ture for the SCF-based impregnation and extraction methodsemployed.

Hilafilcon B SCLs are cross-linked polymeric materials compris-ing the co-monomers represented in Fig. 1, which can easily estab-

Fig. 3. Flurbiprofen loaded in/removed from in Hilafilcon B SCLs after consecutive drug loading/removal cycles. (A) Extracted flurbiprofen amounts (per mg wet SCLs)obtained for SCLs loaded by immersion and leaching in water (h) (small graph) and by scCO2 SSI followed by scCO2 SFE ( ). (B) Flurbiprofen released (per mg wet SCLs) after8 h in water from SCLs loaded by immersion in water (h) (small graph) or that underwent successive SSI/SFE cycles ( ).

1024 F. Yañez et al. / Acta Biomaterialia 7 (2011) 1019–1030

lish hydrogen bonding with flurbiprofen and water, as well asestablish favourable phenyl/methyl/carbonyl/fluorine–scCO2 inter-actions [36,38,39,41,45–48,62–64]. The intense hydrogen bondingbetween ibuprofen and polyvinylpyrrolidone (PVP) after scCO2

impregnation, as well as the ability of scCO2 molecules to interactwith the Lewis basic-type carbonyl group of PVP, have been previ-ously reported [65]. These interactions are also plausible in thecase of flurbiprofen. The relative magnitude of scCO2–flurbiprofenand flurbiprofen–water swollen SCL interactions should controlflurbiprofen ‘‘solubility” in the swollen SCL, as well as its partitioncoefficient between the SCL and the scCO2 phase. If the scCO2–flurbiprofen interactions are less favourable than the flurbipro-fen–water swollen SCL interactions, then flurbiprofen will have ahigher partition coefficient in the SCL, which is a positive aspectfor the SSI process. Otherwise, the drug would be easily removedfrom the SCLs during SFE or during depressurization. Regardingthe scCO2–water swollen SCL interactions, scCO2 readily dissolvesin the SCLs and promotes additional swelling, which may behelpful for both the drug impregnation and extraction processes.Nevertheless, the scCO2–water and flurbiprofen–water interac-tions may also play important roles in these processes [46–48].

One of the great advantages of SCF-based processes is that therelative magnitudes of the scCO2–flurbiprofen and scCO2–waterswollen SCL interactions can be easily tuned, simply by regulatingthe operational pressure, temperature and processing time. Forexample, under the conditions employed in this work (40 �C at12 and 20 MPa) the density of the scCO2 phase (and consequentlyits solvent power) is lower at 12 MPa than at 20 MPa. This meansthat the scCO2–flurbiprofen interactions are favoured at higherpressures. However, the transport properties are also affected: at20 MPa the scCO2 viscosity is greater while its diffusivity is lowerthan at 12 MPa. This means that the transport properties and theplasticization/swelling effects are favoured at lower pressures (be-cause of the enhanced scCO2–water swollen SCL interactions).Thus, if we intend a process in which the partition of a drug be-tween a SCL and the scCO2 phase is fairly high and in which plas-ticization/swelling effects and the potential chain rearrangementscaused by these effects are significant (such as the SSI process em-ployed in this work), then we should choose to work at relativelylow pressures. In contrast, if we intend a process in which the sol-ubility of a drug in the scCO2 phase needs to be improved and inwhich plasticization/swelling effects are less (such as the SFE pro-

cess), then we should work at relatively high pressures. Underthese conditions the transport properties are not favoured and thusthe processing time has to be increased in order to compensate forthis drawback.

In sum, the advantages of loading SCLs with flurbiprofen usingimpregnation with scCO2 compared with immersion in an aqueoussolution can be summarized as follows. Firstly, it is possible to dis-solve more flurbiprofen in scCO2 than in water (this is also true forother low polarity hydrophobic drugs). Secondly, SCF-based pro-cesses can be easily ‘‘tuned” in order to improve drug loading,the drug partition coefficients and the polymer swelling/plasticiza-tion effects that regulate the impregnation and extraction yields.Furthermore, physical rearrangement of the polymeric chains toenable better hosting of the drug molecules may be possible.Thirdly, SCF-based methods are remarkably fast.

Progressive enhancement of the amount of drug impregnated(Fig. 3A) suggests that consecutive flurbiprofen SSI incorporationand SFE extraction steps result in conformational changes in thepolymeric network that create cavities into which the drug canfit. If the conformation of these cavities is stable enough to bememorized in the polymeric network when the drug is extracted,the ability of SCLs to reload flurbiprofen should be greater than be-fore processing. In fact, such an improvement may be further in-creased after successive reload/release cycles as more cavitiescould be generated.

On the other hand, if the affinity of the drug for the polymer in-creases as these tailored cavities increase in number, flurbiprofenrelease should be more sustained. The amount of flurbiprofen re-leased over 8 h in water (Fig. 3B) was remarkably lower (roughlyone-tenth) of that loaded by SSI (Fig. 3A). Although there is an in-crease in the amount released as more SSI/SFE processing cyclesare applied, the ratio between the amount released and that loadedwas not constant but decreased from the one cycle to the three cy-cles processed SCLs (Table 1). The full drug release profiles in waterare shown in Fig. 4. SCLs loaded by the SSI/SFE method sustainedrelease for �3–4 h, while those loaded by immersion in aqueousflurbiprofen solution released significantly lower amounts and sus-tained release for only �1–2 h. Interestingly, increasing the immer-sion time in the flurbiprofen solution from 2.5 to 13 h did not leadto either greater amounts of loaded or released drug (Fig. 5). Re-lease experiments were carried out under near sink conditions,since the intrinsic solubility of flurbiprofen in water is 12 mg l�1

F. Yañez et al. / Acta Biomaterialia 7 (2011) 1019–1030 1025

at 25 �C [59] and the maximum concentration achieved in the re-lease medium was �1.9 mg l�1.

As expected, the scCO2 processed SCLs released greater amountsof drug than those loaded in aqueous flurbiprofen solution,although the percentage released was lower for the former lenses(Table 1). It should be noted that SCLs released less flurbiprofenin 8 h (�54–80%, Table 1) than they could take up in 2.5 h in water.This indicates that Hilafilcon B SCLs have an affinity for flurbipro-fen. However, the lower percentages released by the SCF-processedSCLs (�6–14%, Table 1) suggest that possible rearrangement of thepolymer network during the successive SSI/SFE steps enables high-er affinity drug hosting. ANOVA analysis showed that there are sig-nificant differences between the results obtained after a differentnumber of processing cycles. For example, for the flurbiprofen re-leased from SCLs after three cycle processing the Tukey test (95%significance) indicated statistically significant P values of 60.004and 60.0002 for the SCF-based and for water-based methods,respectively.

Release profiles fitted to Eq. (2) (Table 2) gave quite similar re-lease exponents for SCF and for water processed lenses(0.6 < n < 0.7), with the exception of the three cycle water pro-cessed SCLs. The exponent values suggest that drug release oc-curred by an anomalous transport type (i.e. the superimpositionof Fickian controlled and swelling controlled release [57]), what-ever the loading/removal method. This means that neither the pro-cess nor the number of cycles apparently alter the drug releasemechanism. On the other hand, the release rate constant (k) de-creased as the number of cycles increased, which may indicate thatthe affinity between SCL and flurbiprofen increases as the lensundergoes processing. Similarly, the flurbiprofen diffusion coeffi-cients D1 and D2, obtained by the short-term (Eq. (3)) and thelong-term release approaches (Eq. (4)), respectively, become smal-ler as the number of consecutive processing cycles (for SCF-basedand for water-based processing) increases. The decrease is particu-larly evident in the case of the SCF-processed lenses (Table 2),which confirms enhancement of the affinity for the drug, probablydue to a molecular imprinting effect.

3.2. Molecular imprinting effect

To gain an insight into the conformational changes undergoneby the SCLs that could produce flurbiprofen-imprinted cavities, aproof of concept was performed comparing the sorption of flurbi-profen with that of ibuprofen and dexamethasone from aqueoussolutions. Flurbiprofen, ibuprofen and dexamethasone share somesimilarities in terms of chemical structure (Fig. 1) and have compa-rable (low) solubilities in water (12, 11 and 115 mg l�1, respec-tively) [59,66]. SCF-processed SCLs (after three SSI and SFEcycles) and non-processed SCLs (non-imprinted control polymers)were immersed in solutions of each drug and after loading the re-lease profiles in water were recorded. SCF-processed SCLs couldtake up more flurbiprofen from an aqueous drug solution and inshorter periods of time than control SCLs (Fig. 6A). Furthermore,flurbiprofen uptake is much more efficient than that of ibuprofenand dexamethasone (Fig. 6B). This is also valid for control SCLs,

Table 1Flurbiprofen loaded by SCLs that underwent SSI or conventional loading by immersion in drelative percentage of the amount loaded.

Cycles SCF-based methods (lg mg�1 wet lens) Release/loaded

Released Loaded Residual

1st cycle 1.28 ± 0.07 10.53 0.013 ± 0.003 11.982nd cycle 2.84 ± 0.18 20.29 0.020 ± 0.002 13.863rd cycle 3.58 ± 0.08 54.54 0.063 ± 0.023 6.48

Mean values ± standard deviation.

which confirms that Hilafilcon B SCLs have a higher structuralaffinity for flurbiprofen than for ibuprofen and dexamethasone.The loading of dexamethasone by SCF-processed SCLs and controlSCLs was lowest, being quite similar for both lenses. The relativelyhigh standard deviations were largely due to difficulties experi-enced with the analytical method employed for such highly diluteaqueous drug solutions. Drug release profiles in water showed theopposite behaviour: the most sustained and slowest drug releasewas by the flurbiprofen-loaded SCLs.

Fitting the drug loading and release profiles to Eqs. (2) and (3) isshown in Table 3. The loading exponents were similar for the threedrugs (0.7 < n < 0.9) for the SCF-processed and control SCLs. Thesame was true for the corresponding release profiles in water,although the exponent values were lower (0.5 < n < 0.6). Slightlylower n values were observed for SCF-processed SCLs comparedwith the control SCLs. Regarding the rate constant (k), sorption offlurbiprofen (and dexamethasone) was faster than that of ibupro-fen. The release rate and the diffusion coefficients (D1) followedthe opposite trend. SCF-processed SCLs showed higher sorptionand lower release diffusion coefficients for flurbiprofen than thoseobtained for dexamethasone and ibuprofen (in that order). Addi-tionally, SCF-processed SCLs also presented higher sorption andlower release diffusion coefficients for flurbiprofen than those forcontrol SCLs.

Because the three drugs were at the same molar concentrationin the loading solutions, the differences in loading and releasecould be related to their affinity for the SCLs, their relative sizes/volumes in solution (hydrodynamic Stokes radius or, as an approx-imation, their solid molar volumes), and the network average meshsize. Since all the SCF-processed SCLs passed through the same pro-cessing steps, all of them should possess the same network averagemesh size. The same is expected to occur with the non-processed(control) SCLs, which did not pass through any processing steps.However, dexamethasone has a higher solid molar volume(268.8 cm3 mol�1), as estimated by the Fedors and by the Immi-rzi–Pirini methods [67–69], than flurbiprofen or ibuprofen (183.8and 182.1 cm3 mol�1, respectively). Such a greater size may leadto some constraints on its diffusion and uptake by SCLs, whichcould explain the low dexamethasone loading.

Flurbiprofen and ibuprofen have quite similar molecularweights and molar volumes. Therefore, their different loadingmay arise from different interactions with the polymeric structureof SCF-processed Hilafilcon B SCLs. Flurbiprofen possesses a fluorineatom in one of the aromatic phenyl groups (a C–F bond) and anadditional phenyl group compared with ibuprofen, which has an–CH2–CH(CH3)–CH3 end group. The co-monomers present inHilafilcon B SCLs have a carbonyl group, which is known to interactstrongly and favourably with C–F groups in many chemical/biochemical systems and is involved in several protein/enzymemolecular biorecognition events [51,52]. This type of non-covalentmultipolar interaction usually presents a very characteristic geom-etry: the electronegative organofluorine atom tends to interactorthogonally with the electrophilic carbonyl group and the C–Fbond approaches the plane of the carbonyl group from an angle ofbetween 100� and 140� [51,52,70,71]. The flurbiprofen-imprinted

rug aqueous solution, and flurbiprofen released into water as the total amount or as a

drug (%) Water-based methods (lg mg�1 wet lens)

Released Loaded Release/loaded drug (%)

0.498 ± 0.002 0.92 ± 0.07 54.270.406 ± 0.004 0.49 ± 0.03 82.390.102 ± 0.003 0.147 ± 0.002 69.58

Fig. 4. Kinetics of flurbiprofen release in water for Hilafilcon B SCLs loaded by immersion and leaching in water (small graph, empty symbols: h, 1 cycle; D, 2 cycles; s, 3cycles) and by scCO2 SSI followed by scCO2 SFE (filled symbols: j, 1 cycle; N, 2 cycles; d, 3 cycles).

Fig. 5. Total amount of flurbiprofen released from SCLs loaded by immersion in aqueous flurbiprofen solution for several predetermined periods of time (from 2.5 to 13 h.).

Table 2Flurbiprofen release kinetic parameters obtained applying Eq. (2) and diffusion coefficients estimated by fitting the first 60% of the release curve to Eq. (3) (D1) and the last 60% ofthe release curve to Eq. (4) (D2).

Process Kinetic parameters Diffusion coefficient (mm2 h�1)

n k R2 D1 (�103) R2 D2 (�103) R2

1st cycle SCF processing 0.6985 0.7626 0.9993 1.56 0.9996 0.80 0.99872nd cycle SCF processing 0.7162 0.6029 0.9930 1.18 0.9967 0.32 0.96933rd cycle SCF processing 0.7269 0.5839 0.9934 0.21 0.9986 0.26 0.98721st cycle water processing 0.6495 0.8801 0.9950 1.24 0.9977 1.14 0.99912nd cycle water processing 0.6187 0.7136 0.9964 0.72 0.9970 0.79 0.99993rd cycle water processing 0.4461 0.4128 0.9968 0.66 0.9971 0.71 0.9892

1026 F. Yañez et al. / Acta Biomaterialia 7 (2011) 1019–1030

SCLs may show such biomimetic recognition distinction betweenflurbiprofen and ibuprofen. Dexamethasone also has a C–F bond,but its large molecular size should hinder penetration into the cav-ities formed to host flurbiprofen.

It is feasible that during the SSI process swelling/plasticizationeffects and the interaction of scCO2 with the carbonyl groups ofthe SCL, and flurbiprofen–SCL interactions and templating(through the C–F/carbonyl group interactions), promote polymericrearrangement and the creation of specific ‘‘cavities” with the abil-ity to chemically and structurally recognize flurbiprofen, i.e. a

molecular imprinting effect. Such imprinted ‘‘cavities” may thenbe responsible for the higher sorption of flurbiprofen from aqueousdrug solutions as well as for the observed more sustained drug re-lease profiles.

3.3. Effect of SCF processing on the SCL features

The SEM micrographs (not presented) of the surfaces and cross-sections of non-processed (control), SCF-processed (three cycles ofSSI/SFE) and water-processed (three cycles of water soaking/water

Fig. 6. Molecular imprinting proof of concept for SCF-processed SCLs. (A) Sorption kinetics from flurbiprofen aqueous solutions. (B) Total sorption from aqueous drugsolutions (after 14 h immersion). , SCF-processed SCLs (MIP); h, non-processed/control SCLs (NIP).

Table 3Kinetic parameters of the loading and release of flurbiprofen, ibuprofen and dexamethasone by non-processed SCLs (NIP) and three cycles processed SCLs (MIP-CO2).

Drug Sample Sorption Release

n k R2 n k R2

Obtained from Eq. (2) and considering the first 60% of the total release curveFlurbiprofen MIP-CO2 0.9048 0.7416 0.9994 0.6633 0.7643 0.9908

NIP 0.7290 0.4124 0.9447 0.5237 0.8797 0.9309Ibuprofen MIP-CO2 0.9238 0.5822 0.9985 0.5573 0.8463 0.9525

NIP 0.9054 0.5940 0.9993 0.4908 0.8205 0.9334Dexamethasone MIP-CO2 0.8195 0.7271 0.9870 0.6381 0.7805 0.9642

NIP 0.8727 0.6495 0.9944 0.5298 0.9112 0.9229

Sorption ReleaseD1 � 103 (mm2 h�1) R2 D1 � 103 (mm2 h�1) R2

Obtained from Eq. (3) and considering the first 60% of the total release curve (D1)Flurbiprofen MIP-CO2 1.34 0.9993 1.22 0.9929

NIP 0.47 0.9269 3.32 0.9864Ibuprofen MIP-CO2 0.36 0.9838 2.11 0.9996

NIP 0.79 0.9980 2.17 0.9901Dexamethasone MIP-CO2 0.84 0.9989 1.73 0.9995

NIP 0.56 0.9990 2.80 0.9999

F. Yañez et al. / Acta Biomaterialia 7 (2011) 1019–1030 1027

leaching) did not show any apparent morphological changes be-tween the processed and non-processed SCLs.

The FTIR-ATR spectra (Fig. 7A) of non-processed SCLs (control)and of SCF-processed SCLs (two cycles of SSI/SFE) did not indicatethe presence of remaining flurbiprofen. It also seems that no rele-vant changes occurred in any of the characteristic structural peakof the SCLs. However, a careful analysis of the typical IR bands ofthe SCL co-monomers (Fig. 7B and C) revealed that the strong peakcorresponding to the C@O stretching vibration (�1720 cm�1) wasslightly shifted from 1716 cm�1 (for non-processed SCLs) to1718 cm�1 (for SCF-processed SCLs). The same also happened forthe –OH stretching vibration (�3300–3500 cm�1): the correspond-ing peak was shifted from 3373 cm�1 (for non-processed SCLs) to3383 cm�1 (for SCF-processed SCLs). Despite a FTIR resolution of2 cm�1, these findings are in agreement with the hypothesis thatusing consecutive flurbiprofen SSI/SFE steps, the combined effectsof SCLs plasticization/swelling and of specific flurbiprofen–SCLinteractions (especially through the C–F/carbonyl groups) mayhave caused some SCLs structural rearrangements to create im-printed cavities.

The glass transition temperature of the dried SCLs (91–92 �C)was not altered by successive SCF processing steps. Non-processedand SCF-processed SCLs showed similar degrees of swelling in

water at equilibrium, 57.01 ± 0.92% and 57.88 ± 1.56%, respec-tively, close to that indicated by the supplier (59 wt.%). These smalldeviations could be due to incomplete water removal (some ‘‘bondwater” may still remain under the drying temperature and timeemployed) prior to SCLs re-swelling in water. The mean transmit-tance values in the 400–900 nm range of control, SSI-processed,and flurbiprofen-loaded in water SCLs were 97.4 ± 0.3%,94.7 ± 0.8%, and 93.7 ± 1.5%, respectively. The oxygen permeability(Dk/L) of non-processed SCLs (14.82 barrer cm�1) and of three cy-cle SCF-processed SCLs (14.79 ± 0.66 barrer cm�1) was also unaf-fected by processing with scCO2.

The water contact angles and surface free energies (calculatedby the extended Fowkes (xF) method) are presented in Table 4.The SCF-based and the water-based methods did not greatly affectthe water contact angles for Hilafilcon B SCLs (63–74�) and the sur-face free energies (26–31 mN m�1). The variability in these resultscan be explained by drying (water loss) of the SCLs during the mea-surements, since the sessile drop method is highly sensitive to sur-face dehydration [48,72]. It is clear that the most importantcontributions to the surface energies obtained derive from the dis-persive and hydrogen bonding components. Therefore, all SCLskeep their wettability properties and their surface characteristicsafter processing. Furthermore, we previously observed that the

Fig. 7. (A) FTIR-ATR spectra of non-processed SCLs (thick black line), of SCF-processed SCLs (thin black line) and of flurbiprofen (thin grey line). (B) FTIR spectra for thestretching vibration of carbonyl (C@O) groups: non-processed SCLs (thick black line); SCF-processed SCLs (thin grey line). (C) FTIR spectra for the stretching vibration ofhydroxyl (–OH) groups: non-processed SCLs (thick black line); SCF-processed SCLs (thin grey line).

Table 4SCL water contact angles and surface free energies (calculated by the extended Fowkes (xF) method).

Process Contact angle, H2O (�) Surface free energy, average ± SD (mN m�1)

rS rDS rP

SH–H

Control 69.6 ± 1.9 33.7 ± 2.8 10.0 ± 4.2 0.15 ± 0.21 23.6 ± 6.71st cycle SCF processing 68.1 ± 3.2 31.0 ± 6.0 15.7 ± 11.8 0.10 ± 0.19 15.2 ± 15.52nd cycle SCF processing 70.8 ± 5.0 26.4 ± 1.2 11.3 ± 4.4 0.11 ± 0.22 15.0 ± 5.53rd cycle SCF processing 64.6 ± 0.6 30.1 ± 0.7 17.7 ± 3.3 2.94 ± 4.15 9.5 ± 6.71st cycle water processing 63.7 ± 6.1 30.6 ± 6.5 7.4 ± 4.2 0.00 ± 0.00 23.9 ± 10.02nd cycle water processing 70.8 ± 2.5 32.9 ± 6.5 6.5 ± 4.5 0.77 ± 1.20 25.6 ± 10.33rd cycle water processing 73.5 ± 0.7 35.4 ± 3.8 5.8 ± 1.0 1.55 ± 2.18 28.0 ± 7.0

rS, surface free energy of the solid; rDS , dispersive component; rP

S , polar component; H–H, hydrogen bond component.

1028 F. Yañez et al. / Acta Biomaterialia 7 (2011) 1019–1030

elastic/storage modulus of SCLs does not change after scCO2

treatment.

4. Conclusions

The supercritical fluid processing of SCLs developed in the pres-ent work enables the loading of preformed SCLs with a specificdrug for sustained release. Using scCO2 as the loading/extractionsolvent it was possible to impregnate daily wear Hilafilcon B com-mercial SCLs with flurbiprofen to a greater extent and more rapidlythan using water-based processes. The application of sequentialflurbiprofen impregnation and extraction steps results in the rear-

rangement of some polymeric regions of the SCLs driven by thecombined effect of scCO2 on polymer swelling/plasticization andof specific flurbiprofen–SCLs interactions (mainly through C–F/car-bonyl interactions). This leads to the creation of effective and spe-cific ‘‘cavities” with the ability to chemically and structurallyrecognize flurbiprofen, i.e. a true molecular imprinting effect. Suchimprinted ‘‘cavities” may be then responsible for the observedhigher flurbiprofen sorption from aqueous solution, as well as forthe more sustained release profiles, compared with structurally re-lated drugs. In addition, the employed SCF-based processes did notalter certain critical functional properties of commercial SCLs.Therefore, SCF-assisted methods may be useful for the preparation

F. Yañez et al. / Acta Biomaterialia 7 (2011) 1019–1030 1029

of combination products for the treatment of ocular diseases usingSCLs as the delivery device, while maintaining their functionalityas corrective of refractive deficiencies. Nevertheless, and to estab-lish the range of applicability of the technique, more tests withother drugs and on other functional properties of SCLs (such aspower, geometry, elastic modulus, and fatigue strength) will beperformed in the near future.

Acknowledgements

F.Y. acknowledges the Ministerio de Ciencia e Innovación ofSpain for a FPI fellowship (SAF2005-01930). M.E.M.B. acknowl-edges the FCT-MCTES for a post-doctoral fellowship (SFRH/BPD/21076/2004). This work was financially supported by the FCT-MCTES under Contract No. PTDC/SAU-FCF/71399/2006 and by theMICINN and FEDER (Nos. SAF2008-01679, PT2009-0038). Theauthors also acknowledge Bausch & Lomb (Portugal) and Mart-Op-tic (Coimbra, Portugal) for supplying the contact lenses for thisstudy.

References

[1] Kearns VR, Williams RL. Drug delivery systems for the eye. Exp Rev Med Dev2009;6:277–90.

[2] Calvo P, Vila-Jato JL, Alonso MJ. Comparative in vitro evaluation of severalcolloidal systems, nanoparticles, nanocapsules, and nanoemulsions, as oculardrug carriers. J Pharmacol Sci 1996;85:530–6.

[3] Le Bourlais C, Acar L, Zia H, Sado PA, Needham T, Leverge R. Ophthalmic drugdelivery systems; recent advances. Prog Retin Eye Res 1998;17:33–58.

[4] Sedlacek J. Possibility of application of eye drugs with the aid of gel-contactlenses. Cesk Oftalmol 1965;21:509–12.

[5] Waltman SR, Kaufman HE. Use of hydrophilic contact lenses to increase ocularpenetration of topical drugs. Invest Ophthalmol Vis Sci 1970;9:250–9.

[6] Jain MR. Drug delivery through contact lenses. Br J Ophthalmol1988;72:150–4.

[7] Lesher GA, Gunderson GG. Continuous drug delivery through the use ofdisposable contact lenses. Optom Vis Sci 1993;70:1012–8.

[8] Hiratani H, Fujiwara A, Tamiya Y, Mizutani Y, Alvarez-Lorenzo C. Ocular releaseof timolol from molecularly imprinted soft contact lenses. Biomaterials2005;26:1293–8.

[9] Gulsen D, Chauhan A. Ophthalmic drug delivery through contact lenses. InvestOphthalmol Vis Sci 2004;45:2342–7.

[10] Gulsen D, Li CC, Chauhan A. Dispersion of DMPC liposomes in contact lensesfor ophthalmic drug delivery. Curr Eye Res 2005;30:1071–80.

[11] Novack GD. Ophthalmic drug delivery: development and regulatoryconsiderations. Clin Pharmacol Ther 2009;85:539–43.

[12] Wajs G, Meslard JC. Release of therapeutic agents from contact lenses. Crit RevTher Drug 1986;2:275–89.

[13] McMahon TT, Zadnik K. Twenty-five years of contact lenses – the impact onthe cornea and ophthalmic practice. Cornea 2000;19:730–40.

[14] Karlgard CCS, Wong NS, Jones LW, Moresoli C. In vitro uptake and releasestudies of ocular pharmaceutical agents by silicon-containing and p-HEMAhydrogel contact lens materials. Int J Pharm 2003;257:141–51.

[15] Alvarez-Lorenzo C, Hiratani H, Concheiro A. Contact lenses for drug delivery.Achieving sustained release with novel systems. Am J Drug Deliv2006;4:131–51.

[16] Xinming L, Yingde C, Lloyd AW, Mikhalovsky SV, Sandeman SR, Howel CA, et al.Polymeric hydrogels for novel contact lens-based ophthalmic drug deliverysystems: a review. Cont Lens Anterior Eye 2008;31:57–64.

[17] Peng CC, Kim J, Chauhan A. Extended delivery of hydrophilic drugs fromsilicone–hydrogel contact lenses containing vitamin E diffusion barriers.Biomaterials 2010;31:4032–47.

[18] Alvarez-Lorenzo C, Yañez F, Barreiro-Iglesias R, Concheiro A. Imprinted softcontact lenses as norfloxacin delivery systems. J Control Release2006;113:236–44.

[19] Alvarez-Lorenzo C, Yañez F, Concheiro A. Ocular drug delivery frommolecularly-imprinted contact lenses. J Drug Deliv Sci Technol2010;20:237–48.

[20] White CJ, Byrne ME. Molecularly imprinted therapeutic contact lenses. ExpertOpin Drug Deliv 2010;7:765–80.

[21] Komiyama M, Takeuchi T, Mukawa T, Asanuma H. Molecular imprinting, fromfundamentals to applications. Weinheim: Wiley-VCH; 2003.

[22] Alvarez-Lorenzo C, Concheiro A. Molecularly imprinted polymers for drugdelivery. J Chromatogr B 2004;804:231–45.

[23] Sellergren B, Allender CJ. Molecularly imprinted polymers: a bridge toadvanced drug delivery. Adv Drug Deliv Rev 2005;57:1733–41.

[24] Cunliffe D, Kirby A, Alexander C. Molecularly imprinted drug delivery systems.Adv Drug Deliv Rev 2005;57:1836–53.

[25] Pilestky S, Turner A. Molecular imprinting of polymers. Georgetown,TX: Landes Bioscience; 2006.

[26] Byrne ME, Park K, Peppas NA. Molecular imprinting within hydrogels. AdvDrug Deliv Rev 2002;54:149–61.

[27] Peppas NA, Huang Y. Polymers and gels as molecular recognition agents.Pharm Res 2002;19:577–8.

[28] Bruggemann O, Haupt K, Ye L, Yilmaz E, Mosbach K. New configurations andapplications of molecularly imprinted polymers. J Chromatogr A2000;889:15–24.

[29] Ye L, Mosbach K. The technique of molecular imprinting – principle, state of theart, and future aspects. J Inclusion Phenom Macrocyclic Chem 2001;41:107–13.

[30] Enoki T, Tanaka K, Watanabe T, Oya T, Sakiyama T, Takeoka Y, et al.Frustrations in polymer conformation in gels and their minimizationthrough molecular imprinting. Phys Rev Lett 2000;85:5000–3.

[31] Hiratani H, Alvarez-Lorenzo C, Chuang J, Guney O, Grosberg AY, Tanaka T.Effect of reversible cross-linker, N,N0-bis(acryloyl)cystamine, on calcium ionadsorption by imprinted gels. Langmuir 2001;17:4431–6.

[32] Alvarez-Lorenzo C, Guney O, Oya T, Sakai Y, Kobayashi M, Enoki T, et al.Polymer gels that memorize elements of molecular conformation.Macromolecules 2000;33:8693–7.

[33] Elvassore N, Kikic I. Pharmaceutical processing with supercritical fluids. In:Bertucco A, Vetter G, editors. High pressure process technology: fundamentalsand applications. Amsterdam: Elsevier Science; 2001. p. 612–25.

[34] McHugh MA, Krukonis VJ. Supercritical fluid extraction. 2nd ed. Boston,MA: Butterworth-Heinemann; 1994.

[35] Beckman E. Supercritical and near-critical CO2 in green chemical synthesis andprocessing. J Supercrit Fluids 2004;28:121–91.

[36] Kazarian SG. Supercritical fluid impregnation of polymers for drug delivery. In:York P, Kompella UB, Shekunov BY, editors. Supercritical fluid technology fordrug product development. New York: Marcel Dekker; 2004.

[37] Davies OR, Lewis AL, Whitaker MJ, Tai H, Shakesheff KM, Howdle SM.Applications of supercritical CO2 in the fabrication of polymer systems fordrug delivery and tissue engineering. Adv Drug Deliv Rev 2008;60:373–87.

[38] Kikic I, Vecchione F. Supercritical impregnation of polymers. Curr Opin SolidState Mater Sci 2003;7:399–405.

[39] Kikic I. Polymer–supercritical fluid interactions. J Supercrit Fluids2009;47:458–65.

[40] Kemmere MF, Meyer T. Supercritical carbon dioxide in polymer reactionengineering. Weinheim: Wiley-VCH; 2005.

[41] Kazarian SG. Polymer processing with supercritical fluids. Polymer Sci Ser C2000;42:78–101.

[42] de Sousa HC, Gil MHM, Leite EOB, Duarte CMM, Duarte ARC. Method forpreparing sustained-release therapeutic ophthalmic articles using compressedfluids for impregnation of drugs. EP Patent No. 1611877A1; 2006.

[43] de Sousa HC, Gil MHM, Leite EOB, Duarte CMM, Duarte ARC. Method forpreparing therapeutic ophthalmic articles using compressed fluids. US PatentNo. US20060008506A1; 2006.

[44] Duarte ARC, Simplício AL, Vega-Gonzalez A, Subra-Paternault P, Coimbra P, GilMH, et al. Impregnation of an intraocular lens for ophthalmic drug delivery.Curr Drug Deliv 2008;5:102–7.

[45] Braga MEM, Vaz MT, Costa HSR, Ferreira EI, Gil MH, Duarte CMM, et al.Supercritical solvent impregnation of ophthalmic drugs on chitosanderivatives. J Supercrit Fluids 2008;44:245–57.

[46] Natu MV, Gil MH, de Sousa HC. Supercritical solvent impregnation of poly(e-caprolactone)/poly(oxyethylene-b-oxypropylene-b-oxyethylene) and poly(e-caprolactone)/poly(ethylene-vinyl acetate) blends for controlled releaseapplications. J Supercrit Fluids 2008;47:93–1027.

[47] Costa VP, Braga MEM, Guerra JP, Duarte ARC, Duarte CMM, Leite EOB, et al.Development of therapeutic contact lenses using a supercritical solventimpregnation method. J Supercrit Fluids 2010;52:306–16.

[48] Costa VP, Braga MEM, Duarte CMM, Alvarez-Lorenzo C, Concheiro A, Gil MH,et al. Anti-glaucoma drug-loaded contact lenses prepared using supercriticalsolvent impregnation. J Supercrit Fluids 2010;53:165–73.

[49] Sellergren B. Polymer- and template-related factors influencing the efficiencyin molecularly imprinted solid-phase extractions. Trends Anal Chem1999;18:164–74.

[50] Duarte ARC, Casimiro T, Aguiar-Ricardo A, Simplício AL, Duarte CMM.Supercritical fluid polymerisation and impregnation of molecularlyimprinted polymers for drug delivery. J Supercrit Fluids 2006;39:102–6.

[51] Paulini R, Muller K, Diederich F. Orthogonal multipolar interactions instructural chemistry and biology. Angew Chem Int Ed 2005;44:1788–805.

[52] Hagmann WK. The many roles for fluorine in medicinal chemistry. J Med Chem2008;51:4359–69.

[53] Duarte ARC, Coimbra P, de Sousa HC, Duarte CMM. Solubility of flurbiprofen insupercritical carbon dioxide. J Chem Eng Data 2004;49:449–52.

[54] Owens DK, Wendt RC. Estimation of the surface free energy of polymers. J ApplPolym Sci 1969;13:1741–7.

[55] Rudawska A, Jacniacka E. Analysis for determining surface free energyuncertainly by the Owens–Wendt method. Int J Adhes Adhes 2009;29:451–7.

[56] Korsmeyer RW, Gurny R, Doelker E, Buri P, Peppas NA. Mechanisms of soluterelease from porous hydrophilic polymers. Int J Pharm 1983;15:25–35.

[57] Ritger PL, Peppas NA. A simple equation for description of solute release II.Fickian and anomalous release from swellable devices. J Control Release1987;5:37–42.

[58] Lin CC, Metters AT. Hydrogels in controlled release formulations: networkdesign and mathematical modelling. Adv Drug Deliv Rev 2006;58:1379–408.

1030 F. Yañez et al. / Acta Biomaterialia 7 (2011) 1019–1030

[59] Anderson BD, Conradi RA. Predictive relationships in the water solubility ofsalts of a nonsteroidal anti-inflammatory drug. J Pharmacol Sci1985;74:815–20.

[60] Perlovich GL, Kurkov SV, Bauer-Brandl A. The difference between partitioningand distribution from a thermodynamic point of view: NSAIDs as an example.Eur J Pharm Sci 2006;27:150–7.

[61] Gupta RB, Shim JJ. Solubility in supercritical carbon dioxide. Boca Raton,FL: CRC Press; 2006.

[62] Condo PD, Sumpter SR, Lee ML, Johnston KP. Partition coefficients andpolymer–solute interaction parameters by inverse supercritical fluidchromatography. Ind Eng Chem Res 1996;35:1115–23.

[63] Xu Q, Chang Y. Complex interactions among additive/supercritical CO2/polymer ternary systems and factors governing the impregnation efficiency.J Appl Polym Sci 2004;93:742–8.

[64] Raveendran P, Ikushima Y, Wallen SL. Polar attributes of supercritical carbondioxide. Acc Chem Res 2005;38:478–85.

[65] Kazarian SG, Martirosyan GG. Spectroscopy of polymer/drug formulationsprocessed with supercritical fluids: in situ ATR-IR Raman study ofimpregnation of ibuprofen into PVP. Int J Pharm 2002;232:81–90.

[66] Yalkowsky SH, Yan H. Handbook of aqueous solubility data: an extensivecompilation of aqueous solubility data for organic compounds extracted fromthe AQUASOL database. Boca Raton, FL: CRC Press; 2003.

[67] Fedors RF. A method for estimating both the solubility parameters and molarvolumes of liquids. Pol Eng Sci 1974;14:147–54.

[68] Lyman WJ, Reehl WF, Rosenblatt DH. Handbook of chemical propertyestimation methods. New York: McGraw-Hill; 1982.

[69] Poling BE, Prausnitz JM, O’Connell JP. The properties of gases and liquids. 5thed. New York: McGraw-Hill; 2001.

[70] Olsen JA, Banner DW, Seiler P, Obst-Sander U, D’Arcy A, Stihle M, et al. Afluorine scan of thrombin inhibitors to map the fluorophilicity/fluorophobicityof an enzyme active site: evidence for C–F/C@O interactions. Angew Chem IntEd 2003;42:2507–11.

[71] Pagliaro M, Ciriminna R. New fluorinated functional materials. J Mater Chem2005;15:4981–91.

[72] Read ML, Morgan PB, Maldonado-Codina C. Measurement errors related tocontact angle analyses of hydrogel and silicone hydrogel contact lenses. JBiomed Mater Res Part B 2009;91:662–8.