http://www.revistadechimie.ro REV.CHIM.(Bucharest)♦ 67♦ No. 8 ♦ 20161566

Application of Statistical Design of Experiments for the Optimizationof Clodronate Loaded Liposomes for Oral Administration

IOANA AILIESEI1, VALENTINA ANUTA1, CONSTANTIN MIRCIOIU1, VICTOR COJOCARU1, ANA MARIA ORBESTEANU1,LUDMILA OTILIA CINTEZA1,2*

1Carol Davila University of Medicine and Pharmacy, Faculty of Pharmacy, 6 Traian Vuia Street, 020956, Bucharest, Romania2 University of Bucharest, Faculty of Chemistry, 4-12 Regina Elisabeta Boulevard, Bucharest, Romania

In this paper we used the design of experiments to study the influence of various factors on the formulationof clodronate loaded liposomes. These variables were phosphatidylcholine to cholesterol ratio, the lipidcomponent to active substance ratio and sonication time. The selected responses were drug encapsulationefficiency (DEE, %) and liposomes size (diameter). The statistical experimental design was used to investigatethe influence of individual and combined effects of three factors on the responses and allowed identificationof the main formulation and the quantification of the interactions between the selected factors.

Keywords: liposome formulation, clodronate, design of experiments, optimization

*email: ocinteza@yahoo.com; Tel.: +40722665077

Formulation of a new dosage form or development of aspecific method for quantification or characterization of adrug involves many experiments and sustained efforts. Inthe past years, there are many struggles in this direction,regarding the economic issue of the research and theimprovement of scientist work according to the rapidmarket changes. Therefore, a statistical model, taking intoaccount only the process inputs and outputs, can bedesigned and implemented in the early studies of newcolloidal vectors development. Full or fractional factorialanalysis of experimental data is widely used to designstatistical models consisting of correlations betweendependent and independent process variables [1-8].

Clodronate, a bisphosphonate compound, is an activesubstance with high solubility, low absorption and rapidelimination, commonly used as intravenous and oraltreatment of osteoporosis and several types of cancer [9-12].

Liposomes are colloidal delivery systems used for theencapsulation of various active drugs [13], lipophilic andhydrophilic ones, including clodronate [14].

The aim of this study was to investigate the influence ofindividual and combined effects of three factors (namelyphosphatidylcholine to cholesterol ratio, the lipidcomponent to active substance ratio and sonication time)on the responses of drug encapsulation efficiency (DEE,%) and liposomes size (diameter) and identify the mainformulation in order to produce stable clodronateliposomes with an appropriate bioavailability after oraladministration. Furthermore, an optimization of the drugentrapment into liposomes was performed usingpolynomial mathematical equations and response surfaceplots [15-17].

The application of three-level experimental designs doesnot appear to have been reported in development andoptimization of bisphosphonates incorporation intoliposomes until now.

There are several methods used for the analysis of therelationship between one or more response variables anda set of quantitative parameters, such as completelyrandomized design (CRD), two-level factorial or fractionalfactorial design, response surface methodology (RSM) andTaguchi’s method. Box-Behnken, a form of RSM, is a

convenient three-factors three-coded level design, requiringless runs than Central composite designs [18,19].

Considering these facts, we decided to apply Box–Behnken experimental design for investigation,characterization and optimization of formulationparameters, affecting the stability, encapsulation efficiencyand appropriate size for oral administration of clodronateliposomes.

Experimental partMaterials

Working standard Clodronate monosodium (PharmaZell- Actavis, India), Phosphatidylcholine from egg yolk (SigmaAldrich, Germany), Cholesterol (PanReac AppliChem,Germany), Chloroform solution (Fluka, Sigma-AldrichChemie GmbH, Germany), a solution of 1.5 mM of Cu(NO3)2trihydrated in nitric acid prepared in the Physical ChemistryDepartment, University of Bucharest, Romania.

Formulation of liposomesLiposomes were synthesised by lipid film hydration

method [20]. First, phosphatidylcholine and cholesterolwere mixed and dissolved in chloroform under continuousagitation and the solvent was removed by rotating the vialon a vacuum pump, at room temperature. The lipid filmresulted on the walls was hydrated with an aqueoussolution (1 mg/mL) of active substance heated at 60-70°Cusing the Stuart Apparatus, UC152, Stirrerhotplate, ceramicplate, Staffordshire, UK. The suspension was kept in anultrasound bath for different period of time, according tothe specification of the design. The obtained liposomesare multi-lamellar vesicles with different sizes, appropriatefor oral administration.

Particle size analysisLiposomes were characterized in means of

granulometric distribution using Dynamic Light Scattering,also known as photon correlation spectroscopy or quasi-elastic light scattering technique [21,22]. The particlesdiameter was measured at room temperature for allpreparations by double dilution of the liposomes, first 1:5in distilled water, and 1:50 secondly. The values of size and

REV.CHIM.(Bucharest)♦ 67♦ No. 8 ♦ 2016 http://www.revistadechimie.ro 1567

zeta potential were registered during the experiments madewith Zetasizer Nano ZS, Malvern, Herrenberg, Germany.

Entrapment efficiencyClodronate can form soluble or insoluble complexes,

because of its affinity for metallic ions existing in solution,as well as other members of bisphosphonates drug class.Given the fact that the resulted complexes have thecapacity to absorb light, based on their chromophoregroups, the ratio of active substance can be evaluated fromvarious formulations using a spectrophotometric method.

In this study, the quantitative determination of clodronateis based on its property to form complexes with Cu(II),which determine a specific absorbance in UV/Vis at a pathlength of λ=236nm [23, 24]. The experiments wereconducted on a Shimadzu UVmini-1240 double beam UV-Visible Spectrophotometer, North America. The examinedsamples showed that an indirect determination ofclodronate is needed because Copper ions from thecomplexing agent interfere with the experiment, byabsorbing light itself.

The drug encapsulation efficiency (DEE) represents theratio between encapsulated drug and the initial amount ofdrug, in means of percents: DEE (%) = (Cencapsulated drug/Cinitial drug) x100, where C denotes the concentration (mg/ml) of the indicated substance.

OptimizationOn the basis of previous results [25], a three-level three-

factorial Box–Behnken experimental design was used tooptimize three parameters, i.e. phosphatidylcholine:cholesterol ratio, the lipid component: active substanceratio and sonication time, affecting the formulation ofclodronate into liposomes. Table 1 shows the factorschosen and settings of factor levels, evenly spaced andcoded for low, medium and high settings, as 1.0 and +1[26].

This design is useful for the optimization of the processby using a smaller number of experimental runs, based on

the construction of second order polynomial models andthe exploration of quadratic response surfaces. The modelhas the following form:

where Y represents the measured response, b are theregression, Xi represent the factors studied, Xi

2 quadraticexpressions of the independent variables, XiXj theinteraction terms between variables, and å is the randomerror.

The responses studied are drug encapsulation efficiency(DEE, %) and liposomes size (diameter), as key parametersof liposomes formulation.

The experimental design, data analysis and quadraticmodel building, along with the optimum experimentalconditions generated, were computed by means of Design-Expert 7.0 software (Stat-Ease Inc., Minneapolis, MN, USA).

Results and discussionsExperiments of Box-Behnken design

The design consists of replicated center points and aset of points lying at the midpoints of each edge of themultidimensional cube that defines the region of interest.Therefore, the three-level three-factorial experimentaldesign consists in a total of 17 experimental runs, shownin table 2.

Regarding different combinations of factors and factorlevels, a considerable difference among size distributionwas obtained. The values are ranged from a minimum of219.84 nm to a maximum of 701.97 nm. Also, the resultsfor DEE were between 69.64 and 81.06%, suggesting theinfluence of the selected variables on the responses.

Analysis of variance (ANOVA)Using multiple linear regression analysis, the

coefficients of the polynomial equations were generated :

Table 1FACTORS AND LEVELS OF BOX-

BEHNNKEN EXPERIMENTAL DESIGN

Table 2VARIABLES AND OBSERVED RESPONSES

IN BOX–BEHNKEN DESIGN FORCLODRONATE LIPOSOMES

http://www.revistadechimie.ro REV.CHIM.(Bucharest)♦ 67♦ No. 8 ♦ 20161568

For estimation of significance of the model, the statisticalanalysis of the model was performed in the form of analysisof variance (ANOVA). A model is considered significant ifthe P-value (significance probability value) is less than 0.05,thus very useful in predicting the effects of the threedifferent level factors on the selected responses. Accordingto table 3 and table 4, the model is significant.

Factors effect on selected responsesThe predicted and observed Pearson correlation

coefficients (R2) values for the above mathematicalregressions were 0.9734 for DEE% and 0.9763 for particlesize respectively, indicating that the model interprets thevariability of experimental data. The real relationshipbetween the parameters chosen in this study could beexplained via this model and it is adequate for predictionwithin the range of experimental variables. The statisticalsignificance of the fitted model was checked by F-test andP-value test.

According to Y1 polynomial equation, only X2 (lipid todrug ratio) was found to be significant on DEE respect toother linear coefficients. X1 and X3 were not statisticallysignificant (0.1365 and 0.7796 respectively). However, allinteraction coefficients (X1X2, X1X3, X2X3) and X1

2 , X2

2

quadratic terms are significant, suggesting nonlinearmixed effects on DEE (table 3). The positive value of X2coefficient indicates a favourable effect on DEE. X1X2 haveunfavourable effect on X2 (negative coefficients), whileX1X3 and X2X3 have a favourable effect (positivecoefficients).

For Y2 polynomial equation (size distribution), all thelinear coefficients (X1, X2 and X3), X2

2 and X32 quadratic

term coefficients and X1X2 interaction coefficients werefound significant. The linear coefficients are favouringlarger particle size, whereas the rest are leading to smallersize of liposomes (table 4).

It is presumed that diameter of the vescicles isinfluenced by the quantity of cholesterol added duringpreparation (it is a material that leads to a more compactand rigid lipid film, hard to be broken into small pieces).Also, the more quantity of lipid added during theexperiment the larger size of the liposomes is obtained.

Sonication time should lead to smaller vescicles,because the vibrations create movement through themolecules, clashes and their dispersion into the suspension.

3D response surface plotsThe response surfaces illustrate the relationship

between the dependent and independent variables, in thethree dimensional space, by keeping one factor constantat its central value (fig.1 and fig.2).

OptimizationAn optimal formulation was developed using the

polynomial equations and the surface responsemethodology. It is characterised by the ratio ofphosphatidyl-choline to cholesterol of 2.83:1, a ratio of lipid

Table 3ESTIMATED REGRESSION MODEL OF RELATIONSHIP BETWEEN

DEE% (Y1) AND INDEPENDENT VARIABLES

Table 4ESTIMATED REGRESSION MODEL OF RELATIONSHIP BETWEEN

PARTICLE SIZE (Y2) AND INDEPENDENT VARIABLES

Fig.2. Response surface plots foreffect of formulation parameters

(X1=Ratio of phosphatidylcholine tocholesterol (w/w), X2=Ratio of lipidto drug (w/w), X3=Sonication time(min)) on liposomes mean particle

size

Fig.1. Response surface plots foreffect of formulation parameters

(X1=Ratio of phosphatidylcholine tocholesterol (w/w), X2=Ratio of lipidto drug (w/w), X3=Sonication time

(min)) on drug encapsulationefficiency (DEE %)

REV.CHIM.(Bucharest)♦ 67♦ No. 8 ♦ 2016 http://www.revistadechimie.ro 1569

to drug of 4:1 and 1 min sonication time as the optimumvalues for the independent variables (desirabilityvalue=0.929), with 80.68 for drug encapsulation efficiencyand 298.241 for size distribution as fitted responses (fig.3).

The results suggested that DEE% is less influenced bythe selected independent variables in comparison withParticle Size, with only drug to lipid ratio showing to besignificant. The findings are also consistent with ourprevious studies [25], where influence of the same selectedfactors was found to be similar on entrapment ofalendronate in liposomes. On the contrary, particle sizewere significantly influenced by the selected factors forboth clodronate and alendronate liposomes. Even thoughthe use of different types of lecithin leads to major changesin particles size values (higher range for egg lecithin respectto soy lecithin), the dimensions of colloidal vescicles arestill appropriate for oral administration.

It was seen that the highest DEE% is attributed toformulations in which the ratio of Lipid to drug is 4:1 (table4) and statistically it is not affected by the phosphati-dylcholine to cholesterol ratio, even though during theexperiments, according to table 2, run sample no.12 ,where the ratio was 5:1 (less cholesterol), this lead to asample with high encapsulation efficiency and run sampleno.13 lead to a sample with lower encapsulationefficiency, in which the ratio was 2:1 (more cholesterol).Nevertheless, the sonication time also influenced DEEvalues, even though it wasn’t statistically significant. Forthe same samples, 1 min of sonication (run no.12) wasenough to allow the encapsulation of more quantity of drug,while 5 min tent to a smaller amount incorporated (runno.13). As it can be seen in the same table, run sampleno.14 expresses the more evident influence of thesonication time respect to run sample no.12, 9 min ofsonication lead to 70.64% respect to 80.21% for 1 minsonication time.

In practice, the statistical design is only a tool to properlyadjust the parameters; the results may be a little bit differentthan the prediction, because the interactions between theselected factors are complex and depend on lab conditionsas well.

To confirm the validity of the calculated optimalparameters and predicted responses, clodronateencapsulation into liposomes at optimal combination offormulation variables was carried out. The suitability of themodel predicting the optimum response values and the

Fig.3. Optimization of clodronate loaded liposomes bymeans of desirability function in ramp function graph

representation. Red dots indicate the optimum values ofthe independent variables (X1=Ratio of

phosphatidylcholine to cholesterol (w/w), X2=Ratio oflipid to drug (w/w), X3=Sonication time (min)) and blue

dots are the predicted values of responses (drugencapsulation efficiency (DEE, %) and liposomes size) at

the optimum factor levels

Table 5OBSERVED AND PREDICTED RESPONSES (Y1, Y2) AND RESIDUAL VALUES FOR THE DRUG INCORPORATION

INTO LIPOSOMES, PERFORMED AT OPTIMAL VALUES OF FACTORS INVESTIGATED IN THE STUDY

results are represented in table 5. The optimizedcombination of investigated preparation conditions ensuredthe synthesis of liposomes, which had the characteristicsvery close to the predicted values.

ConclusionsThe synthesis of clodronate disodium loaded liposomes

with optimal characteristics was established usingexperimental design methodology. Box-Behnken designoffered the possibility of analyzing different models for drugdelivery system with clodronate, which revealed higherentrapment efficacy and theoretical bioavailability of theactive substance than the market products. The results ofthe study pointed out that lipid to drug ratio was thepredominant factor that influenced drug encapsulationefficiency and liposomes size distribution were affectedby all the experimental conditions. The optimumformulation suggested by the design is defined by a ratio ofphosphatidylcholine to cholesterol of 2.83:1, a ratio of lipidto drug of 4:1 and 1 minute sonication time. By using thisapproach, it is possible to create the appropriateexperimental conditions, as an economic way to producean efficient and stable formulation for the clodronateliposomes.

Acknowledgement. The work has been funded by the SectoralOperational Programme Human Resources Development 2007-2013of the Ministry of European Funds through the Financial AgreementPOSDRU/159/1.5/S/132395.

References1.YUANGYAI, C., NEMBHARD, H.B., Emerging Nanotechnologies forManufacturing (Second Edition), A volume in Micro and NanoTechnologies, 2015, p230–2542.KUMAR, S., XU, X., GOKHALE, R., BURGESS D.J., InternationalJournal of Pharmaceutics, 464, nr.1–2, 2014, p34–453.GUL, T., BISCHOFF, R., PERMENTIER, H.P., Electrochimica Acta,171, 2015, p23–284.TAJBER, L., CORRIGAN, O.I., HEALY, A.M., International Journal ofPharmaceutics, 367, nr. 1–2, 2009, p86–965.KUMAR, N., GOINDI, S., International Journal of Pharmaceutics,472, Nr. 1–2, 2014, p.224–2406.MAHESHWARI, V., RANGAIAH, G.P., LAUB, T., SAMAVEDHAM, L.,Chemical Engineering Science, 131, 2015, p84–907.KUMAR, L., REDDY S. M, MANAGULI, R. S., GIRISH PAI K., SaudiPharmaceutical Journal, 2015, http://dx.doi.org/10.1016/j.jsps.2015.02.001

http://www.revistadechimie.ro REV.CHIM.(Bucharest)♦ 67♦ No. 8 ♦ 20161570

8.ARUS, V.A., NISTOR, I.D., PLATON, N., ROSU, A.M., MUNTIANU, G.,JINESCU, C., REV. CHIM. (Bucharest), 66, nr. 1, 2015, p889.DAVISON, N.L., GAMBLIN, A., LAYROLLE, P., YUAN, H., DE BRUIJN,J., BARRERE-DE GROOT, F., Biomaterials, 35, nr. 19, 2014, p5088–509710.ZHU, J., ZHENG, Y., ZHOU, Z. et al, Bone, 67, 2014, p189–19211.ZHU, J., ZHENG, Y., ZHOU, Z., European Journal of Cancer 49, 2013,p2086– 209212.MACHADO, M., CRUZ, L.S., TANNUS, G., FONSECA, M., ClinicalTherapeutics/Volume, 31, nr. 5, 2009, p962-97913.TORCHILIN V.P., Recent advances with liposomes as pharmaceuticalcarriers, Nat. Rev. Drug Discov., 4, nr. 2, 2005, p145-16014.LOVE, W.G., CAMILLERI, J.P, WILLIAMS, B.D., Journal ofPharmacological and Toxicological Methods, 27, nr. 3, 1992, p185–18915.ANDERSON, M.J., WHITCOMB, P.J., Productivity Press, New York,200516.DAVID, S., MARCHAIS, H., BEDIN D., CHOURPA, I., InternationalJournal of Pharmaceutics, 478, nr. 1, 2015, p409–41517.KINCL, M., TURK, S., VRECER, F., International Journal ofPharmaceutics, 291, 2005, p39–49

18.RAGONESE, R., MACKA, M., HUGHES, J., PETOCZ, P., J. Pharm.Biomed. Anal., 27, 2002, p995–100719.BOX, GEP, BEHNKEN, D.W., Technometrics, 2, 1960, p455–47520.VAN ROOIJEN, N., SANDERS, A., Journal of Immunological Methods174, 1994, p83–9321.BERNE, B.J., PECORA, R., Dynamic Light Scattering; Wiley: NewYork, 197622.LI, Z., WANG, Y., SHEN, J., LIU, W., SUN, X., Optics and Lasers inEngineering, 56, 2014, p94–9823.KOBA, M., KOBA, K., PRZYBOROWSKI, L., Acta PoloniaePharmaceutica-Drug Research, 65, nr. 3, 2008, p289-29424.WALASH, M. I., METWALLY, M. E.-S., EID, M., EL-SHAHENY, R. N.,International journal of Biomedical science, 4, nr. 4, 2008, p303-30925.AILIESEI, I., CINTEZA, L.I., ORBESTEANU, A.M., COJOCARU, V.,ANUTA, V., Studia Universitatis Vasile Goldis, Seria Stiintele Vietii, 24,nr. 1, 2014, p73-7926.SINGH, K.S., DODGE, J., DURRANI, M.J., KHAN, M., Int. J. Pharm.,125, 1995, p243–255

Manuscript received: 10.07.2015