Comparison of in vitro tests at various levels of complexity for the prediction of in vivo performance of lipid-based formulations: Case studies with fenofibrate

  Research paper Comparison of   in vitro  tests at various levels of complexityfor the prediction of   in vivo  performance of lipid-based formulations:Case studies with fenofibrate Brendan T. Griffin a, ⇑ , Martin Kuentz b , Maria Vertzoni c , Edmund S. Kostewicz d , Yang Fei d , Waleed Faisal a ,Cordula Stillhart b , Caitriona M. O’Driscoll a , Christos Reppas c , Jennifer B. Dressman d a School of Pharmacy, University College Cork, Ireland b University of Applied Sciences and Arts Northwestern Switzerland, Institute of Pharma Technology, Muttenz, Switzerland c Faculty of Pharmacy, National & Kapodistrian University of Athens, Greece d Institut für Pharmazeutische Technologie, Goethe Universität, Frankfurt am Main, Germany a r t i c l e i n f o  Article history: Received 22 July 2013Accepted 25 October 2013Available online 31 October 2013 Keywords: Lipid-based formulationsPoorly soluble drugsFenofibratePrecipitation In vitro  digestion In vitro  dispersion testing In vivo  bioavailabilitySolid state characterisation a b s t r a c t The objectives of this study were to characterise three prototype fenofibrate lipid-based formulationsusing a range of   in vitro  tests with differing levels of complexity and to assess the extent to which thesemethods provide additional insight into  in vivo  findings. Three self-emulsifying drug delivery systems(SEDDS) were prepared: a long chain (LC) Type IIIA SEDDS, a medium chain (MC) Type IIIA SEDDS, anda Type IIIB/IV SEDDS containing surfactants only (SO). Dilution, dispersion and digestion tests were per-formed to assess solubilisation and precipitation behaviour  in vitro . Focussed beam reflectance measure-ments and solid state characterisation of the precipitate was conducted. Oral bioavailability wasevaluatedin landrace pigs. Dilution anddispersion testing revealed that all three formulations were sim-ilar in terms of maintaining fenofibrate in a solubilised state on dispersion in biorelevant media. During in vitro  digestion, theTypeIIIAformulationsdisplayedlimiteddrugprecipitation(<5%),whereastheTypeIIIB/IV formulation displayed extensive drug precipitation (  70% dose). Solid state analysis confirmedthatprecipitatedfenofibratewascrystalline.Theoralbioavailabilitywassimilarforthethreelipidformu-lations (65–72%). Insummary, the use of LC versus MC triglycerides in Type IIIASEDDS had no impact onthe bioavailability of fenofibrate. The extensive precipitation observed with the Type IIIB/IV formulationduring  in vitro  digestion did not adversely impact fenofibrate bioavailability  in vivo , relative to the TypeIIIA formulations. These results were predicted suitably using  in vitro  dilution and dispersion testing,whereas the  in vitro  digestion method failed to predict the outcome of the  in vivo  study.   2013 Elsevier B.V. All rights reserved. 1. Introduction Lipid-basedformulations(LBF)haveemergedasausefulformu-lation strategy to increase the bioavailability of poorly water solu-ble drugs [1,2]. The design of LBF, and in particular appropriate excipient selection to ensure a predictable  in vivo  performance,can be challenging. Choosing excipients can involve a balance be-tween achieving complete drug solubility in the formulation, andmaintainingthedruginasolubilisedstatethroughoutthedilution,dispersion and digestion processes which occur in the gastrointes-tinal tract (GIT). The impact of these processes on  in vivo  perfor-mance is still poorly understood and there is a considerable riskof drug precipitation during the various stages of dilution, disper-sion and digestion [3,4]. Drug precipitation  in vivo  may pose a riskto drug absorption, as reversion to a crystalline form and subse-quent need to undergo re-dissolution, negates the advantage of formulating a poorly water soluble drug in a solubilised state.Importantly, the extent of precipitation appears to be formulationdependentbuttodateourabilitytocorrectlypredictthelikelihoodof precipitation, and impact thereof, is limited.There is a need in the industry for the availability of rapid char-acterisation techniques that would facilitate early phase screeningof LBF in pre-clinical development. Guidance from ICH Q8 and theQuality by Design (QbD) initiatives establish a need for morereliable  in vitro  tests than mere screening of potentially poor for-mulation candidates. Therefore, further research is needed to de-velop predictive  in vitro  techniques for the early and advancedstage characterisation of LBF. The three most common testscurrently used to evaluate the performance of LBF are the  in vitro dilution test, dynamic dispersion test and  in vitro  digestion test(Table 1). 0939-6411/$ - see front matter   2013 Elsevier B.V. All rights reserved.http://dx.doi.org/10.1016/j.ejpb.2013.10.016 ⇑ Corresponding author. Tel.: +353 (0) 21 4901657; fax: +353 (0) 21 4901656. E-mail address:  brendan.griffin@ucc.ie (B.T. Griffin).European Journal of Pharmaceutics and Biopharmaceutics 86 (2014) 427–437 Contents lists available at ScienceDirect European Journal of Pharmaceutics and Biopharmaceutics journal homepage: www.elsevier.com/locate/ejpb  The simplest  in vitro  test can be performed by diluting the for-mulation in appropriate media. This is a typical screening tool foran early development phase, because it allows miniaturisationand parallel testing of formulation candidates. In-line techniquessuch as focussed beam reflectance measurement (FBRM) can becombined with dilution testing to detect drug precipitation [5,6].The dynamic dispersion test, using the USP2 apparatus providesan assessment of the solubilisation behaviour of the formulationunder compendial hydrodynamic conditions and, in addition, canfacilitate a kinetic assessment of precipitation behaviour. Whensuch dynamic dispersion tests are run in biorelevant media, theycan be useful in the prediction of the  in vivo  performance of LBFviaan in vitro – in silico – in vivo  approach[7,8]. The in vitro  digestionmethod is the most widely referenced technique for assessing theimpact of lipolysis on the formulation [9,10]. However, this tech-nique has some drawbacks in terms of requiring extensive samplepreparation to quantify drug content of the various digest phases;it is not readily amenable to high throughput analysis and despiterecent progress in terms of optimising the test conditions, arrivingat adequate prediction of   in vivo  performance has proven elusive[11]. In summary therefore, the three tests differ in their degreeof complexity, usefulness at different stages in discovery/develop-ment, amenability to high throughput screening and most impor-tantly relevance to the  in vivo  situation (Table 1).The current study was designed to characterise the dispersioncharacteristics of the different LBF using the three most common in vitro  methods and to relate the results with  in vivo  bioavailabil-ity in pig. Fenofibrate (a BCS class II compound), was chosen as amodellipophilicdrug(log P   =5.24, Sol=0.3mg/l at37  C). Thebio-availability of fenofibrate from conventional dosage forms is low,variable and increased when administered after food [12,13] andtherefore possesses characteristics that may benefit from formula-tion as a lipid-based system.Three prototypes LBF were developed and classified in accor-dance with the Lipid Formulation Classification System (LFCS)[14]. The LFCS represents a scheme for categorisation of diverse li-pid-based drug delivery systems according to their compositionand properties. The LFCS classifies the various types of LBF fromsimple drug loaded oils with one excipient (Type I) to self-emulsi-fying systems (Type II), self-microemulsifying systems which con-tain a significant proportion of oils (Type IIIA) and those which arepredominantly water-soluble surfactants (Type IIIB), to pure sur-factant only systems (Type IV). The three formulations were pre-pared using the surfactants Cremophor RH (HLB  15) and Tween85(HLB  11) at afixedratio1:2. TwoType IIIAformulationsweredesigned, containingeither40%mediumchaintriglyceride(MC) or40% long chain triglyceride (LC), which were then designated MCType IIIA and LC Type IIIA respectively. The third formulationwas a surfactant-only (SO) formulation (i.e. triglyceride free),which was classified as Type IIIB/IV. All three produced finenano-emulsions on dispersion in water. The key rationale behindselectionofthesethreeformulationswastofacilitateacomparisonbetween (a) MCT versus LCT as the lipid component and (b) diges-tiondependent(TypeIIIA)versusdigestionindependent(TypeIIIB/IV)formulationson in vitro and in vivo characteristics.Additionally,these potential differences are being evaluated in the absence of co-solvents in the formulation. Co-solvents are frequently addedtotheformulationtoenhancethesolubilityofthedrugintheSED-DS. However, when attempting to assess the risk of precipitationusing  in vitro  tests, the presence of co-solvent in the dilution/dis-persion media may impact the likelihood for precipitation, partic-ularly for methods where non-sink conditions apply. In summary,theobjectivesofthisstudyweretoevaluatethebehaviourofthreeprototype LBF using a range of standardised  in vitro  tests withincreasing levels of complexity and to assess the extent to whichthese methods may provide insight into  in vivo  findings, particu-larly in relation to addressing the risk of drug precipitation. 2. Materials and methods  2.1. Materials Olive Oil ‘highly refined low acidity’ (C 18  triglycerides) andTween85(Polyoxyethylene-(20)–polysorbitantrioleate)werepur-chased from Sigma–Aldrich (Ireland). Miglyol 812 (M812)(Caprylic/Capric (C 8 –C 10 ) triglycerides) was received from SasolGermany. Cremophor RH 40 (Polyoxyl-40-hydrogenated castoroil) was received from BASF (Germany). Porcine pancreatin(4xUSP)andsodiumtaurocholate(>95%)werepurchasedfromSig-ma–Aldrich (Ireland). Fenofibrate was purchased from Sigma Al-drich Chemie GmbH, Germany. Hard Gelatine Capsules (Size 0)were obtained from Capsugel (Coni-Snap  ). Lecithin (Lipoid E PCS, >98% pure) was kindly donated by Lipoid GmbH (Ludwigshafen,Germany). All other chemicals and solvents were of analyticalgrade or HPLC grade respectively.  2.2. Preparation of SEDDS formulations Ternary phase diagrams were established for both the LCT (Ol-ive oil/Cremophor RH/Tween 85) and MCT (Migylol/CremophorRH/Tween 85) systems. Subsequent screening of the self-emulsify-ing properties was conducted by establishing the pseudo-ternaryphase diagram after dispersion (1:100) in water, using a fixed sur-factant: co-surfactant (Cremophor RH: Tween 85) ratio of 1:2. ALCT-SEDDS (40% LCT), a MCT SEDDS (40% MCT) and a surfactant-aloneSEDDS(i.e. noTG)wereselected,representingaLC-TypeIIIA,MC Type IIIA and a Type IIIB/IV formulation, respectively. Each  Table 1 Comparison of the three  in vitro  characterisation techniques in terms of relevance and usefulness for prediction of   in vivo  performance of lipid based formulations. Dispersion DigestionDilution test a Dynamic dispersion test Lipolysis testExperimental simplicity ++ +   Proximity to  in vivo  situation   /+ b + ++Miniaturisation for early formulation screening ++ +   Suitability for testing of the final dosage form (e.g. capsule)    ++   c Quantification of the extent of precipitation   /+ b ++ ++Determination of supersaturation and kinetics of precipitation   /+ b ++ ++Solid state analysis of precipitate   /+ b ++ ++Relevance ranking: ++=high, +=medium,   =limited. a Macroscopic and microscopic evaluation. b Dependent on experimental scale. c Limited physiological relevance.428  B.T. Griffin et al./European Journal of Pharmaceutics and Biopharmaceutics 86 (2014) 427–437   formulation was prepared by adding excipients into a screw capglassvial, followedbyvortexingtoensurecompletemixing.Fenof-ibrate solubility in each SEDDS was determined as previously de-scribed [15]. For  in vitro  and  in vivo  studies, fenofibrate wasadded to the lipid blend (80mg/g) and mixed at 50  C for  30min, followed by overnight storage at 37  C. Thereafter, theformulations were stored at ambient temperature prior to evalua-tion. For  in vivo  studies, the formulationswere handfilled intotwohard gelatinecapsules (2  600mg) on the day prior to dosing andstored at ambient temperature.  2.3. Supersaturation and droplet size following dilution/dispersion The solubility of fenofibrate following dispersion of the SEDDS1:5 and 1:200 in SGF sp  was determined initially. Equilibrium solu-bility was determined by preparing drug-free systems (i.e. SEDDSdiluted1ml in5ml and0.03ml in6ml), addinganexcess amountof fenofibrate to the mixtures and stirring (500rpm) at 37  C inhermeticallysealedglassvials.After24h,48h,and72h,analiquotwas withdrawn, centrifuged (14,000rpm, 37  C, 30min) and theconcentration of drug was determined in the supernatant usingHPLC. Equilibrium was assumed when two consecutive solubilitysamples varied by  6 5% (w/w). Experiments were conducted intriplicate. The solubility values at each dilution were then usedto calculate the maximum supersaturation ratio ( SR M  ) as previ-ously described [16] SR M  ¼ Maximumtheoreticalconcentrationof fenofibrateinmediaFenofibratesolubilityfollowingdispersioninmedia  ð 1 Þ SR M  is the ratio of the fenofibrate dose in the formulation (i.e. themaximum theoretical concentration of solubilised drug in the dilu-tion media in the absence of any drug precipitation) and the drugsolubility in the dilution media. A  SR M  <1 indicates that the formu-lation is below saturation, whereas a value  SR M  >1 represents asupersaturateddrugstate.Dropletsizeandpolydispersityindexfol-lowing1:200dilutionweredeterminedbyphotoncorrelationspec-troscopy (PCS) using Malvern Zetasizer instrument (Nano-ZS, UK).  2.4. Dispersion experiments The dilution and dispersion experiments were conducted usingbiorelevant media under conditions simulating the pre-prandialstomach using either FaSSGF or FaSSGF-v2 [17,18] and under con-ditions simulatingthe pre-prandial small intestine using FaSSIF-v2[17].  2.4.1. Dilution in biorelevant media – microscopic and macroscopic evaluation Thedispersioncharacteristicsofthedifferentlipidformulationsof fenofibrate wereevaluatedinFaSSGF andFaSSIF-V2at a1:5and1:200 dilution ratio. The diluted samples were vortexed to ensurecomplete dispersion and stored at 37  C. At pre-determined inter-vals up to 24h, the samples were investigated both microscopi-cally and at the macroscopic level. The mixtures were examinedmacroscopically without prior shaking [image capture with a dig-ital camera (Lumix DC Vario, Panasonic, Germany)] and comparedwith the drug-free samples. Microscopic evaluations of the sam-ples were undertaken at 30min and 120min to enable qualitativeidentification of crystal formation. Samples were withdrawn fromthe diluted formulation (which had been gently shaken by handfor approximately 2s), placed onto a glass microscope slide whichwas then overlaid by a coverslip and viewed at 40  magnification(Leica DMI 400B).  2.4.2. Dynamic dispersion Kinetic assessment of dispersion and precipitation characteris-ticswasperformedusingthemini-paddleapparatus(Erwekamod-ified DT6 dissolution system, Heusenstamm, Germany), in water,SGF sp  (USP 30), FaSSGF-V2 and FaSSIF-V2. The geometry of theused model is a scaled-down version of the USP2 apparatus (pad-dle) so that hydrodynamic conditions remain essentially similar[19]. The tests were performed in 200ml medium per vessel at37  C±0.5  C and paddle revolution speed was 75rpm, using twogelatine capsules (containing in total 96mg of fenofibrate), thatwere positioned into the vessel using sinkers. Sampling was per-formed prior to and 5, 10, 15, 20, 25, 30, 40, 60, 90 and 120minafter the initiation of the experiment, using a 5ml glass syringe(Fortuna  Optima  ) connected with stainless steel cannula. Ateach time point, 4ml of the mediumwas withdrawn and immedi-atelyfilteredthrougha 0.45 l mpore sizeregeneratedcellulosefil-ter (Titan 2, 17mm, SunSri, USA), discarding the first 2ml. Thewithdrawn volumewas replaced withfresh, pre-warmed medium.Filtrates were diluted with mobile phase and analysed by HPLC.Experiments were performed in triplicate.  2.5. Focused beam reflectance measurement  Focused beam reflectance measurement (FBRM) was used todetect precipitation of drug particles after dispersion of the threeformulations with and without fenofibrate (80mg/g) [20]. The for- mulations were evaluated following dilution 1:5 and 1:200 inSGF sp  and in water. The Lasentec FBRM D600L probe (Lasentec,Redmond, USA) was positioned at a 30   angle in a thermostatedglass vessel (37  C), and the dispersion was stirred at 400rpm.Measurements were recorded every 5s for 2h using the iC FBRMsoftware version 4.0 (Mettler-Toledo AutoChem, Columbia, USA).Experiments were performed in triplicate.  2.6. In vitro lipolysisIn vitro  lipolysis experiments were conducted using a methodsimilar to that previously described [21]. Briefly, 1g of LBF was dispersed in 90ml of digestion buffer (50mM tris maleate,150mM NaCl, 5mM CaCl 2 , pH=7.5) containing 5mM taurocholicacid and 1.25mM phosphatidyl-choline (conditions broadly rep-resentative of fasted state intestinal conditions). Digestion wasinitiated by the addition of 10ml of pancreatin extract, freshlyprepared as previously described [21], to the 90ml of digestion media to produce a final lipase concentration of 1000 tributyrinunits (TBU) per ml of digest media. Lipolysis was followed over80min at 37  C using a pH-stat titration unit (718 STAT Titrino,Metrohm Ltd., Switzerland), which maintained the pH at 7.5using 0.2M NaOH. Blank digestion experiments (i.e. no formula-tion) were performed to account for background fatty acid pro-duction derived from the digest media. Aliquots of 4.9ml werewithdrawn from the digestion medium at time zero (before addi-tion of lipase solution) and at pre-determined time intervals afteraddition of the lipase solution. Immediately after sample with-drawal, 9 l l/ml of 0.5M solution of p-bromophenyl boronic acidin methanol was added to quench further lipolysis. Samples weresubsequently centrifuged using a Beckman L7-56 ultracentrifuge(Beckman Instruments, Palo Alto, USA) equipped with NVT90 ro-tor for 90min at 37  C and 205,422  g   (54,000rpm) in order to sep-arate the sample into three phases: an intermediate aqueousphase (containing bile salts, fatty acids and MG), an upper oilyphase (containing undigested TG and/or DG) and a pellet phase.Each layer was then diluted with acetonitrile and assayed fordrug content using HPLC. B.T. Griffin et al./European Journal of Pharmaceutics and Biopharmaceutics 86 (2014) 427–437   429   2.7. X-ray diffraction and Raman spectroscopy X-ray diffraction (XRD) and Raman spectroscopy were used tocharacterise the solid-state of any drug particles that evolved dur-ing dispersion and lipolysis testing. The dispersion test was al-lowed to proceed for 2h, after which any solid particles werecollected by filtration. The XRD patterns and Raman spectra wererecorded immediately following sample preparation. As a refer-ence, the XRD and Raman results of untreated crystalline fenofi-brate was also acquired. Samples from the pellet phase collectedafter ultra-centrifugation (i.e. 80min post  in vitro  lipolysis) werealso evaluated. As a reference the pellet phase obtained from thelipolysis of drug-free formulation was spiked with crystallinefenofibrate. The XRD was performed using a X-ray powder diffrac-tometer (R-XRD Phaser D2, Bruker AXS GmbH, Karlsruhe, Ger-many) equipped with a Co and Cu KFL tube (30kV, 10mA) asradiation source and a Lynxeye  detector. A single sample wasscanned in the angular range of 5  (2 h ) to 40  (2 h ) with a step sizeof 0.1   (2 h ) and a count time of 5s per step. For Raman spectros-copy,aRamanRXN1analyzer(KaiserOpticalSystems,Inc.,AnnAr-bor, MI) in the backscattering mode was utilised, equipped with adiode laser operating at a wavelength of 785nm (laser power400mW). A multi-fibre P h AT probe having a non-contact samplingdevice with a laser spot diameter of 6mm was used. Spectra werecollected at a resolution of 4cm  1 and an acquisition time of 5susing the iC Raman Instrument software (Version 3.0, Mettler-Toledo AutoChem Inc., Columbia, MD).  2.8. In vivo bioavailability study All experimental procedures were approved and performed inaccordance with licences issued by the Department of Health, Ire-land as directed by the Cruelty to Animals Act Ireland and EU Stat-utory Instructions. Six male Landrace pigs (15–20kg, mean17.5kg) were fasted for 16h before experiments. On day 1, anindwelling intravenous (i.v.) catheter was inserted into the jugularvein, under general anaesthesia as previously described [22]. Onday 3, following an overnight fast, the oral formulations wereadministered in two gelatine capsules (containing in total 96mgoffenofibrate)withtheaidofadosinggun, afterwhichthepigsre-ceived  50mloftapwaterviaasyringe.Tofacilitatehandlingdur-ing oral dosing, a mild relaxant intramuscular dose of ketamine(5mg/kg) and xylazine (1mg/kg) was administered. After dosing,pigs were returned to their pens and blood samples (4ml) werecollected at time zero (pre-dosing) and 0.5, 1, 2, 3, 4, 6, 8, 12, 24and 48h post-dosing. Water was available ad libitum throughoutthe study period and the animals were fed 8h post-dose. All bloodsamples were collected in heparinised tubes (Sarstedt, Germany)andimmediatelycentrifugedat 4000rpmfor 5minat 4  C. Plasmawas collected and stored at  80  C until analysis by HPLC.The study was conducted as a three way randomised crossover,with a 7day washout period. However, two of the indwelling ve-nous catheters became dislodged on the last leg of the study. Asa result, data were obtained from 5 animals following dosing of either Type III A MC and Type IIIB/IV formulations and from 6 ani-mals following dosing of the Type IIIA LC formulation. For the i.v.dosing, a separate group of 4 pigs were administered 25mg fenof-ibrate dissolved in ethanol-based solution [23]. The i.v. solution(3ml) containing 8.33mg/ml fenofibrate in 80% w/w ethanol(96%w/v)+20% saline was administered by slow infusion over2min directly into an ear vein. Blood sampling was performed asoutlined previously for the oral formulations, with an additional3bloodsamplestakenat0.0833,0.25,0.75hpost-dose.Allanimalsremained in good health throughout the study.  2.9. Quantitative analysis of fenofibrate The concentrations of fenofibrate in samples from solubilitystudies and dissolution/dispersion tests were determined by HPLCusing previously described methods [24,25]. The pharmacokinetic evaluationof fenofibratewas based onthe quantificationof fenofi-bric acid, the major active metabolite of fenofibrate, using a vali-dated HPLC-UV method, based on previous published methodswith minor modifications [12,26]. 0.5ml plasma was spiked with internal standard (Sulindac) followed by addition of 0.5ml of 25% NaCl solution and 1ml of 1% H 3 PO 4  in methanol. The samplewas mixed thoroughly and subsequently centrifuged. 20 l l of theclearsupernatantswasinjectedontoaSynergi,C18reversedphaseHPLC column (250  4.6  2.6 l m; UV detection at 286nm; mo-bilephase 80%methanol: 20%water (pHwas adjustedto 2.5), flowrate 1ml/min). The analysis of fenofibric acid in plasma displayedexcellent linearity ( r  2 P 0.99) over the concentration range of 50–2000ng/mlandtheLOQwas80ng/ml.Theextractionrecoveriesof fenofibric acid and internal standard were >95%. The precision of the method at 50, 400 and 1600ng/ml expressed as the coefficientof variationwas 7.8%, 4%and 0.9% within days and 12.5%, 3.9%and0.8% between days, respectively.  2.10. Pharmacokinetic analysis The peak plasma concentrations ( C  max ) and the time for theiroccurrence ( T  max ) were noted directly from the individual plasmaconcentration versus time profiles. The elimination rate constant( k el )wasestimatedbyregressionanalysisfromtheslopeofthelineof best fit of the log plasma concentration time curves, and thehalf-life( t  1/2 )ofthedrugwasobtainedby0.693/ k el . Theareaunderthe plasma concentration versus time profiles from 0 to 48h(AUC 0 ? 48 ) was calculated by the linear trapezoidal method.(AUC 0 ? 1 ) was calculated by addition of (AUC 0 ? 48 ) to the extrapo-lated tail area calculated using the 48h plasma concentration di-vided by  k el . The absolute bioavailability ( F  a ) was calculated as: F  a  ¼  AUC  oral  AUC  i : m :     Dose i : m : Dose oral   A comparison of pharmacokinetic parameters was performed usingANOVA following Tukey’s post-test at a significance level of  a  =0.05. All statistical analysis was performed using IBM  SPSS  Version 20. 3. Results  3.1. Formulation characteristics A summary of the self-emulsification properties and solubilityof fenofibrate in the three SEDDS is presented in Table 2. The sur-factant: co-surfactant ratio and dose-load of fenofibrate (80mg/g)was held constant among the three formulations, which repre-sented a saturation level of 56%, 83% and 77% for the Type IIIAMC, Type IIIA LC and Type IIIB/IV formulations, respectively. The SR M  values following dilution of the SEDDS in SGF sp  (at 1:5 and1:200 dilution) are also presented. Following dilution of the TypeIIIA MC formulation, the  SR M  was <1, indicating that the formula-tion was below saturation at each dilution ratio. In contrast, TypeIIIA LC became supersaturated on dilution in SGF, and the effect ismore pronounced at low levels of dilution. The Type IIIB/IV for-mulation displayed the highest  SR M  on dilution, which wouldindicate the greatest risk of drug precipitation from thisformulation. 430  B.T. Griffin et al./European Journal of Pharmaceutics and Biopharmaceutics 86 (2014) 427–437    3.2. Dilution in biorelevant media Macroscopicandmicroscopicevaluationwasperformedfollow-ing dilution in bio-relevant media (FaSSGF and FaSSIF-V2) (Fig. 1).The three formulations formed isotropically clear, transparentnano-emulsions on dilution 1:200 in either FaSSGF or FaSSIF-V2.The Type IIIB/IV formulation formed a slightly opalescent colloidaldispersion, which had a bluish white appearance, indicative of alarger droplet size, which is also supported by PCS data (Table 2).At the lower level of dilution, the dispersions were less transpar-ent, with the degree of turbidity being greater in FaSSGF media.Microscopically, the Type IIIAformulations exhibitedminimal pre-cipitation. For the Type IIIB/IV, threadlike particles were observedat the1:5dilutionratio, whileat the 1:200ratio, theoverall extentof precipitation was low. While precipitation was qualitativelygreatestatthe1:5dilutionratio,thisrepresentstheworstcasesce-nario in vivo ,inwhichtheformulationwouldundergolowlevelsof dilutionwithinthe GI tract for a short periodof time. At the higherdilution level, precipitation was minimal, which would reflect thebest case scenario in which the formulation is administered to-gether with at least 200ml of fluid. In the media simulating thefasting small intestine (FaSSIF-V2), a similar trend was observedforeachformulationwithsomeprecipitationobservedfortheTypeIIIB/IV formation at low dilutions levels while at a 1:200 dilutionthe degree of precipitation was minimal and qualitatively similarfor all three formulations.  3.3. Focused beam reflectance measurement  FBRM was applied to analyse the appearance of drug particleswhich precipitated during dispersion of the three formulations inSGF sp  or in water [20]. This technique consists of an in-processprobe, which measures the chord length distribution of evolvingparticles. Only counts with a chord length of less than 10 l m wereselectedto minimise the noise resulting fromaggregated particles.However, it should be noted that the detection is limited to parti-cles with a chord length of more than  1 l m. For the Type IIIA MCand LC formulations, there was no significant increase in particlecounts upon formulation dispersion, relative to the drug-free for-mulation (data not shown). In contrast, there were considerablyhigher FBRM counts measured from dispersion of the Type IIIB/IVformulation, at both dilution ratios. However, at a lower dilutionratioof 1:5w/wanotable signal noise was observeddue toforma-tion of droplets, which interfered with the signal arising from pre-cipitating particles (Fig. 2a and c). At the higher dilution ratio of 1:200 w/w, this signal noise of the formulation was lower. Here,the number of chord lengths clearly increased after 30min of dis-persion(Fig. 2b andd). This highdilutionlevel enabledmonitoring  Table 2 Composition, and corresponding solubility, droplet size and maximum supersaturation ratios ( SR M  ) after dilution in SGF sp  of the three SEDDS. Formulation Composition % w/v Fenofibrate solubility(mg/ml)Dropletsize (nm)Poly-dispersityindexSaturation levelin SEDDS SR M  at different dilutionlevels1:5 1:200Type IIIA MC 40% Miglyol 20% Cremophor RH 40% Tween 85 143.8±10.9 72.3±0.3 0.18±0.01 0.56±0.04 0.86±0.03 0.74±0.02Type IIIA LC 40% Miglyol 20% Cremophor RH 40% Tween 85 96.6±3.4 51.7±0.8 0.12±0.00 0.83±0.03 1.55±0.14 1.22±0.03Type IIIB/IV 33% Cremophor RH 67% Tween 85 104.4±7.7 159.4±0.9 0.27±0.01 0.77±0.06 3.41±0.13 3.90±0.30 (a)(b) Fig. 1.  Macroscopic and microscopic observations following dilution of the lipid-based formulation in biorelevant media. A type III A MC, Type IIIA MC and a Type IIIB/IVformulation were diluted 1:5 and 1:200 in FaSSGF and FaSSIF-V2. (a) Macroscopic observations of glass vials containing drug-free (left) and fenofibrate-loaded formulations(right) under various dilution conditions. (b) Micrographs showing images of fenofibrate precipitates observed to varying degrees in the media,   120min after dilution. (Forinterpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) B.T. Griffin et al./European Journal of Pharmaceutics and Biopharmaceutics 86 (2014) 427–437   431