Hydroxy fatty acid based polyanhydride as drug delivery system: Synthesis, characterization, in vitro degradation, drug release, and biocompatibility

Hydroxy fatty acid based polyanhydride as drug deliverysystem: Synthesis, characterization, in vitro degradation,drug release, and biocompatibility

Jay Prakash Jain, Sweta Modi, Neeraj KumarDepartment of Pharmaceutics, National Institute of Pharmaceutical Education and Research, Sector 67,SAS Nagar, Mohali 160062, India

Received 19 June 2006; revised 20 March 2007; accepted 9 April 2007Published online 16 July 2007 in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/jbm.a.31456

Abstract: Low molecular weight hydroxy fatty acidbased polyanhydrides were synthesized by one potmethod, a variable of typical melt-condensation and char-acterized by FTIR, NMR, DSC, and GPC. Polymerdegrades by both surface and bulk erosion as trailed byweight loss, anhydride loss and surface morphology.Control over drug release was accessed with drugs fea-turing different aqueous solubility, that is, methotrexate(hydrophobic) and 5-fluorouracil (hydrophilic). Effect ofloading, at 5, 10, and 20% w/w of methotrexate onrelease profiles was also studied and negligible effect wasdiscovered. Biocompatibility of polymers was evaluatedin SD rats after SC injection of the polymer. Histopathol-

ogy revealed initial inflammation of the tissues near theinjection site however healed with time. Overall, thesepolymers were found good to control the release of theentrapped drug and were found biocompatible in prelim-inary in vivo study. Due to their low melting tempera-tures they can be injected locally (SC or intratumorally) tofrom regional in situ depot and have a great potential asa drug carrier for localized delivery of anticancer drugs.� 2007 Wiley Periodicals, Inc. J Biomed Mater Res 84A:740–752, 2008

Key words: biodegradable polymers; fatty acid polyanhy-drides; drug delivery; methotrexate; biocompatibility

INTRODUCTION

Various types of polyanhydrides have been usedsince 1980’s as drug carrier in different diseased con-ditions.1,2 For a particular application the rate ofdegradation of polyanhydrides can be manipulatedby change in the monomer type and ratio withoutaffecting their attribute of biodegradability. Thus,selection of monomers becomes crucial to obtain thepolymer of desired degradability. Fatty acids arefound good candidates for the synthesis of biode-gradable polymers as they are naturally occurringhydrophobic components.3,4 Besides affecting thedegradation pattern these also affect the physicalstate of the final polymer, which in turn govern theiruse either as solid implantable material or directlyinjectable carrier systems. Owing to the importanceof anhydride bond in the controlled drug delivery

application, different type of fatty acid polyanhy-dride biomaterials have been assessed as potentialcarrier for bioactive agent.3,5,6 The major limitationin synthesizing polymer containing fatty acid is thedearth of difunctionality of fatty acid so that it canbe integrated in the polymer chain.

Ricinoleic acid (RA or cis-12-hydroxyoctadeca-9-eonoic acid), which is the one of few commerciallyavailable hydroxy fatty acid, has been found to be themost appropriate alternative for the synthesis of thefatty acid based polyanhydrides. Its advantage lies inbifunctionalilty due to a 12-hydroxyl group alongwith the acid group and, therefore, can be incorpo-rated into the polyanhydride backbone by the forma-tion of an ester bond. Previous reports on polyanhy-dride synthesized from RA maleate or succinate andSA (sebacic acid) demonstrated the hydrophobicity,flexibility, biocompatibility and biodegradability4 butthese polyanhydrides were high molecular weightblock-co-polymers having high melting point frombody temperature perspective, hence difficult toinject. Moreover, Shen et al. have suggested thatblock-co-polymers undergoes microphase separationdue to differences in relative hydrophobicity of thecomonomers, resulting in thermodynamic partition-

Correspondence to: N. Kumar; e-mail: [email protected] grant sponsors: Third Word Academy of Sci-

ence (TWAS), NIPER; contract grant number: NIPER Com-munication No. 398

' 2007 Wiley Periodicals, Inc.

ing of drugs incorporated into these copolymers.7

Sikanov et al.8 inserted the RA in a preformed poly(sebacic acid) chain, which involved synthesis of pre-polymer of SA followed by insertion of RA into thepolymer chain and resulted in block copolymer. Withthis background, the aim of this study was to synthe-size an injectable low molecular weight polymer withrandom distribution of RA and SA monomers in onestep. Polymers were synthesized by a method vari-able of melt-polycondensation, and have been eval-uated for degradation, in vitro drug release specifi-cally using hydrophobic/hydrophilic drugs with theeffect of loading level, and biocompatibility.

MATERIALS AND METHODS

Methotrexate and 5-fluorouracil were kind gift samplesfrom Astron Pharmaceuticals (Ahmedabad, India) and Bio-chem Pharmaceutical Industries (Mumbai, India), respec-tively. Castor oil and RA were purchased from sigma (Ger-many) and Hi-Media (India), respectively. Maleic anhydridewas obtained from SD fine chemicals (India) and acetic an-hydride from Qualigens (India). All other reagents exceptHPLC solvents were of analytical grade from Lobachemie(India) or SD fine chemicals (India) and used as received.The HPLC solvents were obtained from Rankem (India).Sprague-Dawely female rats of 250–300 g were obtainedfrom Central Animal Facility, NIPER (Mohali, India)

Instrumentation

Infrared (IR) spectra of the samples were obtained usingPerkin–Elmer (USA) spectrometer. Solid samples were ei-ther pressed into KBr pellets or cast onto NaCl plates, whilesemisolid samples were dissolved in chloroform for record-ing IR Spectra. Samples for NMR were prepared by dissolv-ing them in CDCl3 and analyzed using Bruker (Avane DPX300) (Germany). Molecular weights of the polymers wereobtained from GPC. Polystyrene standards in the molecularweight range of 682 Da-54 kDa were used as reference.HPLC grade chloroform was used as mobile phase for bothpolymer and standards. GPC was carried out at ambienttemperature on a divinyl benzene column (10 A) (PolymerStandard Services, Warwick) at a flow rate of 1 mL/minwith injection volume of 50 lL. The effluent was monitoredon the UV–visible detector attached to the HPLC system atwavelength (kmax) of 254 nm. The data was analyzed using‘‘GPC for Class-VP’’ software. Melting point of the polymerswas determined by DSC (Mettler Toledo, Switzerland) atheating rate of 108C/min and hot stage microscopy (HSM)(Leica, Germany). Surface morphology of the device duringdegradation was studied by scanning electron microscopy(SEM) (VP-420, LEO electron microscopy Ltd., England).The samples were mounted on the stub and coated withgold, using sputter coater. Electron microscopy of the coatedsamples was performed and surface images at various lev-els of magnifications were captured. Methotrexate and 5-flu-orouracil were analyzed by validated HPLC methods usingC18 reverse-phase (5 lm) column (Lichrospher 100 RP-18e)

(Merck, Germany). Mobile Phase for methotrexate consist-ing phosphate buffer (0.01M, 5.75 pH) (Solution A) and[Methanol þ acetonitrile (11:5)] (Solution B) in the propor-tion of 86:14 was used as eluent at a flow rate of 1 mL/minand the effluent was monitored at 303 nm using UV detec-tor. For 5-fluorouracil, phosphate buffer (0.01M, 5.75 pH)and methanol in the proportion of 95:5 was used as eluentat a flow rate of 0.6 mL/min with UV detection at 266 nm.

Synthesis of RA and SA copolymer

Separation of fatty acids from castor oil

Castor oil was refluxed with 0.5M ethanolic KOH at1108C for 30 min, reaction mixture was then acidified with1N HCl. Fatty acid was separated in chloroform and evapo-rated to dryness. Purification was performed by columnchromatography using cyclohexane:ethyl acetate (9:2) as elu-ent. The locating agents were bromocresol green solutionand iodine vapor. Purified product was stored at �208C.

Synthesis of ricinoleic acid maleate

RA obtained from castor oil was used for the synthesisof ricinoleic acid maleate (RAM) with the reportedmethod.4 Briefly, RA and maleic anhydride (1:2 molarratio) in toluene were taken and stirred overnight at 908C.After completion of the reaction, reaction mixture waswashed four times with distilled water. The product wasthen dried over anhydrous Na2SO4 (1:1 v/w) for an hour.Column chromatography of the product obtained was per-formed on a silica gel column (silica mesh 60–120) withpetroleum ether/ethyl acetate/acetic acid (80/30/1). TLCof the fractions obtained was carried out to confirm thedesired product.9 The final product obtained was a clear,oily, and yellowish color liquid, which was characterizedby FTIR and NMR.

FTIR spectra of RAM has shown peak at 2929.5 and2857.0 cm�1 corresponding to C��H stretching bend, whilepeak at 1710 cm�1 was for carbonyl group. 1H NMR of RAMdemonstrated the characteristic peaks at particular chemicalshifts (d), which include 6.36 (dd, 2H, HOOC��CH¼¼CH��COO), 5.43 (m, 1H, CH2��CH¼¼CH��CH2), 5.26 (m, 1H,CH2��CH¼¼CH��CH2), 5.02 (quintet, 1H, OCO��CH), 2.31(m, 2H, CH2��COOH), 2.26 (m, 2H, ��CH¼¼CH), 2.01 (m, 2H,HC¼¼CH��CH2��CH2), 1.62 (m, 2H, CH2��CH2��COOH),1.50 (m, 2H, OCO��CH��CH2��CH2), 1.29 (m, 16H, 8CH2s),0.92 (t, 3H,��CH3). It clearly indicated the formation of RAM.

Copolymerization of various ratios of RAmaleate and SA

To get homogenously distributed RA and SA monomerunit in the polymer one-pot melt-polycondensationmethod was used. RAM and SA both are bifunctional withtwo carboxylic acid groups in their structure. Differentratios of RAM and SA (10:90, 30:70, 40:60, 50:50, 60:40,70:30, 90:10) were taken in a round bottom flask andrefluxed with acetic anhydride (1:5 w/v) at 1508C for

HYDROXY FATTY ACID BASED POLYANHYDRIDE AS DRUG DELIVERY SYSTEM 741

Journal of Biomedical Materials Research Part A DOI 10.1002/jbm.a

30 min. Excess acetic anhydride was evaporated at 408Cunder vacuum. This product was subjected to high vac-uum (<4 3 10�3 mbar) at 658C for 90 min with continuousN2 flushing after every 15 min for 30 s.

A separate polymer by typical melt condensationmethod was synthesized to evaluate microphase separa-tion in block-co-polymer. Briefly, prepolymers of both themonomers viz. RAM and SA were prepared by refluxingthem with acetic anhydride (5 v/w) for half an hour andexcess acetic anhydride was evaporated in the rotavapor.The prepolymers (1:1 w/w) were melt-condensed similarto the condensation procedure of random copolymer(658C, 90 min), so that the effect of low synthesis tempera-ture or low molecular weight on phase separation can alsobe evaluated against the typical melt-condensation process,which is carried out at 150–1808C.

The polymers were characterized by FTIR, NMR, GPC,DSC, and HSM. Anhydride bond showed typical doubletasymmetric and symmetric stretch bend in FTIR at 1820.7and 1781.1 cm�1, respectively, while ester group appearedas strong stretch peak at 1752.9 cm�1. Additionally theC��O stretch shows four more bends in the range of 1300–1000 cm�1 and one of these bends was broader thanothers. All these bends indicate the presence of ester andanhydride bond in the structure and hence polymerization.1H NMR has shown characteristic peaks at different chem-ical shift (d) values as: 6.88 (dd, 2H in head-to-head RAM,CO��CH¼¼CH��CO), 6.28 (d, 2H in head-to-tail RAM,CO��CH¼¼CH��CO), 5.46 (m, 1H, O��CO��CH2��CH¼¼CH��CH2), 5.35 (m, 1H, O��CO��CH2��CH¼¼CH��CH2), 5.00 (quintet, 1H, methane), 2.52 (t, 6H, CH2��COO),2.26 (m, 2H, O��CH��CH2��CH¼¼CH), 2.00 (m, 2H,O��CH��CH2��CH¼¼CH��CH2), 1.60 (m, 6H, CH2��CH��COOCO), 1.40–1.60 (m, 2H, OCO��CH��CH2��CH2), 1.3(m, 24H, (CH2)12, from RAM and from SA), 0.90 (t, 3H,CH2��CH3). The peak at d 6.28 illustrate the presence of��CH¼¼CH�� group flanked by two anhydride bonds,while this peak in RAM shows dd at d 6.88 because ofasymmetric environment.

Drug loading and fabrication of the delivery device

The devices were fabricated by melt casting method. Aparticular amount of each of the polymer was taken in a5-mL-beaker and melted just above its melting point in awater-bath. In case of loaded device the weighed drug (5, 10,or 20% w/w) was mixed in the melt, while the blank poly-mer devices were cast as such. After mixing the melt waspoured into the custom-made molds of silicone rubber toobtain homogeneous size of the device. The size of individ-ual slot in the mold was 11 3 5 3 3 mm3. The devices wereweighed and used for further studies. The same procedurewas followed for loading and device casting with 5-FU.

In vitro degradation studies

Wet and dry weight loss

The blank cast devices of representative polymer with50:50 ratios of RAM:SA, were weighed and placed sepa-

rately in the flat bottom glass vials with capacity of 30 mL.Twenty milliliter of phosphate buffer (0.1M, 7.4 pH) wasadded to each vial and kept in the shaker water bath at378C and 100 rpm. At each sampling point, the polymersamples were taken out of the buffer and gently washedthree times with 5 mL of distilled water. The wet weightof the devices was taken and samples were then freezedried overnight to obtain the dry weight. Sampling wasdestructive and experiment was carried out in triplicatefor each time point.

Anhydride loss

The polymer contains hydrolysable anhydride linkages,which cleave during polymer degradation. The freezedried samples of different time point were taken from theweight loss degradation study and IR spectra of the samewere taken. The anhydride loss was assessed using FTIRby means of determining area of, an asymmetric peak ofanhydride, which appears at around 1820 cm�1 and anacid peak at 1700 cm�1. The percent anhydride loss wascalculated by the Eqs. (1) and (2) given below.

Peak ratio ¼ m1820 ðAnhydrideÞm1820 ðAnhydrideÞ þ m1700 ðAcidÞ ð1Þ

% Anhydride loss

¼ Peak ratio of polymer at specified time 3 100

Peak ratio of polymer at zero timeð2Þ

Surface morphology

Surface morphology of the device during degradationwas studied by SEM. Surface morphology was observedon intact polymer devices and method for the same isdescribed in instrumentation section.

In vitro drug release studies

The weighed drug-incorporated devices were placed in30-mL glass vial and 20 mL of the phosphate buffer (0.1M,7.4 pH) was added to each container. The release studieswere carried out at 378C with stirring at 100 RPM. Sampleswere withdrawn at different time intervals with thereplacement of equal amount of the release media to main-tain the sink condition. All the samples were filteredthrough the 0.45-lm mesh nylon filters and analyzed inHPLC after appropriate dilutions. A set of standard sam-ples were also analyzed on the same day to obtain the cali-bration curve. The release studies were conducted in simi-lar manner for devices with 10 and 20% MTX loading, anddevices loaded with 5-FU.

Drug release kinetics

Kinetics of drug release from polymer devices isdescribed using zero-order, first-order, Higuchian and Hix-

742 JAIN, MODI, AND KUMAR


son–Crowell’s models. Zero-order kinetics10 show linearrelationship between amount released and the time [Eq.(3)]. Constant release rates can be achieved with systemsfollowing this model. First-order kinetics [Eq. (4)] describesrelease kinetics of a controlled release system when therelease of drug is proportional to the amount of drugremaining to be released. Higuchi described drug dissolu-tion from matrix based modified release systems as a dif-fusion process and based on Ficks’s law, squire root timedependent expression was derived [Eq. (5)].11,12 Hixson–Crowell (1993) described a model [Eq. (6)], assuming therelease rate to be limited by the drug particles dissolutionrate and not by the diffusion that might occur through thepolymeric matrix.13 This model recognizes that the particleregular area is proportional to the cubic root of its volumeand can be applied in devices where surface diminishesduring the dissolution.

Qt ¼ K0t ð3Þ

ln Qt ¼ ln Q/ þ K1t ð4Þ

Qt ¼ KHt1=2 ð5Þ

Q/1=3 �Qt1=3 ¼ KHCt ð6Þ

where, Qt is the amount of drug released in time t, Q! isthe initial amount of drug and K0, K1, KH, and KHC arerelease rate constants for zero order, first order, Higuchiand Hixson–Crowell equation, respectively. Dissolutiondata was fitted to these models and regression analysiswas carried out. The criterion for selecting the most appro-priate model was based on best goodness-of-fit.

In vivo studies

Biocompatibility of the polymer

Sprague–Dawely rats (250–300 g) were used to examinethe biocompatibility of the polymer. Anesthesia, surgicaland injection procedures were justified in detail and wereapproved by Institutional Animal Ethics Committee (IAEC,NIPER). All the animals were housed individually in plas-tic cages in a controlled environment (22–248C and 12:12light/dark cycle) with free access to food and water.

Twenty five Animals were divided in three groups, inwhich one group served as control (n ¼ 1). In the othertwo groups, 50:50 RAM:SA (n ¼ 2) and 70:30 RAM:SA(n ¼ 2) polymer were tested for biocompatibility. Thepolymer samples, RAM:SA 50:50 and 70:30, were gentlywarmed to 45 and 408C, respectively. 0.5 mL of the liquidpolymer samples were injected subcutaneously with 18-Gauge needle in the rat back at neck region and controlgroup was injected with 0.5 mL of normal saline. Individ-ual animal weight was taken daily, prior to their sched-uled sacrifice. Animals were sacrificed at 1, 3, 7, 12, and22 days following injection. The tissue in the 1 cm areafrom the injection site was removed and fixed by placingit into 10% formal saline for 48 h. The tissue was regularly

processed for light microscopy. All the tissues were firstdehydrated by placing them in gradient increasing concen-tration of absolute alcohol and xylene. The anhydrous tis-sue samples were then embedded in paraffin blocks. Fivemicrometer-thick-sections of the tissue were cut with thehelp of microtome and processed for hydration and stain-ing was performed with hematoxylin-eosin dye.

Inflammation level in each section was given gradefrom 0 to 4 on the basis of acute–chronic inflammatorysymptoms, fibroblastic proliferation, collagen formation,presence of foreign body, or any other symptoms ofinflammation. The overall response was determined byexamining the section considering the above attributeswhere 0 means absence; 1 means minimal; 2 means moder-ate; 3 means moderate to severe; and 4 means severe formof inflammation.

RESULTS AND DISCUSSION

Synthesis of RA and SA copolymer

Castor beans obtained from Ricinus communis con-tain 50–55% of castor oil. The oil itself contains anumber of fatty acids like RA, oleic acid, linoleic acid,stearic acid, and palmitic acid, however, among allthese fatty acids, RA content is *90%.14 Glycerides incastor oil were hydrolyzed by saponification and thefatty acid phase was separated with chloroform fol-lowed by evaporation under vacuum to obtain thefatty acids. The yield of fatty acids was 75%.

Most of the fatty acids like stearic acid, palmiticacid, and so forth, which occur naturally are mono-functional without double bond and can only be usedto cease the preformed polymer chain and providethe hydrophobicity to the final polymer.15 Some fattyacids like erucic acid and oleic acid contain doublebond in their structure and hence can be converted tofatty acid dimer or trimer having 2 or 3 carboxylicgroup, respectively, for further polymerization.16,17

RA is the only fatty acid which is commercially avail-able and contains two functional groups, that is, onecarboxylic and one hydroxyl group.18 12-Hydroxygroup of the RA can be transformed to carboxylicgroup by esterification so that both carboxylic groupscan be utilized for homogeneous polymerization.Esterification of RA can be carried out with maleic an-hydride or succinic anhydride, but maleic anhydridewas preferred, due to one additional double bondwhich decreases the melting point of the polymer andfacilitate injection.4 The fatty acids obtained from cas-tor oil was reacted overnight with maleic anhydridein 1:2 molar ratios at 908C in toluene, the final reactionmixture contains around 15% of fatty acids and theiresters other than RAM. These 15% monoacids act aschain terminators and hinder the polymerization reac-tion. To overcome this problem, virtually pure diacid



monomer was obtained by column chromatography,using petroleum ether/ethyl acetate/acetic acid (80/30/1) as eluent. The concentrated fractions whichhave shown single spot in TLC9 corresponding toRAM were pooled and the final product obtained wasa clear, oily, and orange color liquid, which wasstored at �208C.

In this study low molecular weight polymer withrandom distribution was synthesized by a variablemethod of polycondensation with various ratios(10:90, 30:70, 40:60, 50:50, 60:40, 70:30, 90:10) of RAMand SA (Scheme 1). With increased percentage ofRAM polymer gradually become semisolid to liquidand polymers with more than 70% of RAM wereessentially thick liquids. The color of the polymerwas also transformed from white to dark orange asa function of RAM content because of its dark or-ange color, while the SA was white powder.

Molecular weight and melting pointof the polymers

The weight average molecular weights along withpolydispersity of all polymers are shown in Table IThe synthesized polymers were of low molecularweight as compared with the molecular weight ofpolymers synthesized by usual melt condensationmethod.4 Low molecular weight of the polymerobtained was due to two main factors, that is, the

starting material and reaction conditions. First factorcontributed in a way that RA used for the synthesisof RAM was separated from castor oil hence it con-tains some of the monofunctional fatty acid, whichcan terminate the chain. Though, RAM was purifiedby column chromatography but the RA used for itssynthesis was obtained form castor oil, while earlierreports have used technical grade RA thus somemonofunctional fatty acid impurity can be expected,

Scheme 1. Complete reaction scheme for synthesis of RA and SA copolymer. Different ratios (10:90, 30:70, 40:60, 50:50,60:40, 70:30, 90:10) of RAM and SA were condensed to form the polymer.

TABLE IMolecular Weight, Polydispersity, and Melting

Temperatures of the Synthesized Polymer

RAM:SA FeedRatio (w/w)

Weight RatioCalculated by

1HNMRaMwb

(Da) PIc Tmd (8C)

10:90 9.6:90.4 3900 2.14 54.1630:70 25.6:74.4 3550 2.08 43.8340:60 42.5:57.5 3700 2.05 42.4050:50 46.6:53.4 3800 1.90 41.0660:40 34.2:65.8 4000 2.50 40.5070:30 70.0:30.0 3400 2.01 31.4690:10 88.1:11.9 3800 2.10 Liquid at RT

aWeight ratio calculated by NMR peaks at chemical shift(d) values 5.46 (O��CO��CH2��CH¼¼CH��CH2) plus 5.35(O��CO��CH2��CH¼¼CH��CH2) versus 1.3 (CH2)12 of bothricinoleic and sebacic acid.

bWeight average molecular weight.cPolydispersity determined by GPC.dMelting temperature (determined by DSC).



which was also evident from very small peak in 1HNMR of RAM at d value of 3.3 ppm. Secondly, thereaction temperature at the time of condensationwas deliberately kept less than the temperature gen-erally used for melt-polycondensation. Thereforethese two major aspects during synthesis renderedthe polymer low molecular weight. Molecular weightof the polymer was calculated by GPC, which didnot show any specific trend in molecular weight andpolydispersity with respect to different ratios ofRAM and SA, the probable reason can be the lowmolecular weight of the polymer. Further, a singlepeak in GPC is confirmation of polymerization andmitigates the chances of blend formation. Meltingbehavior of the polymers was studied by DSC andHSM (Table I). Since SA is a crystalline material19 itshigher ratio in the polymer results in higher Tm,while RA, a thick liquid, cause decline in Tm.

Randomization of RAM and SA component in thecopolymer was observed in DSC thermogram. Forthis purpose, block-co-polymer with molecularweight of 6400 Da was synthesized. Differentialscanning calorimetry (DSC) was used to identifymicrophase separation20–22 and a comparison wasmade for both the polymer, that is, random andblock-co-polymer, containing 50:50 ratios of RAMand SA. DSC results indicated that Both polymershave two different peak shape (Fig. 1). A single peak[Fig. 1(a)] was observed for random polymer con-firming random distribution of monomers, while dis-torted peak shape was observed for block copoly-mer, which is a merger of two peaks [Fig. 1(b)]; firstit is showing the endothermic peak for RAM partthen the peak of SA (SA have crystallinity of 66%).This suggests the random nature of disclosed poly-mers.

In vitro degradation studies

Polymers were evaluated for weight loss, anhy-dride loss, and surface morphology during degrada-tion. Weight loss was conducted to observe the timecourse of degradation and water uptake by the de-vice. Degradation studies were carried out in phos-phate buffer (0.1M, pH 7.4). During the course ofdegradation the device loose around 80% of itsweight in 10 days, while at the same time point de-vice comprised of 34% water (Fig. 2). Swelling (swel-ling was considered when the weight of the deviceincreases from immediate previous time point) inthe polymer device was not observed though thewater uptake by the device increase with time andthe difference between the wet and dry weightdecreased. This behavior can be explained byincreased in porosity of the device with time so thatmore and more amount of water penetrates into thedevice. The net weight loss imitates first orderbehavior while wet weight loss comport zero orderdegradation probably due to water penetration intodevice was proportional to the porosity and hencetime (Fig. 2, inset).

SEM was used to study the process of degradationby exploring the surface morphology of the polymer.SEM pictures of degrading polymer at different timeintervals revealed the increase in porosity with time(Fig. 3) and it was in agreement to the stated weightloss study. Initially an integrated net of structurewas seen but gradual widening of the meshworkcaused overall increase in porosity of the device,which allows more water to penetrate the device,thus the difference between dry and wet weightdecreased with time.

Although polyanhydride degrades primarily bysurface erosion, there are many factors, which influ-

Figure 1. Differential scanning thermogram of RAM:SA50:50 (a) random copolymer, and (b) block copolymer.Clearly showing the microphase separation in block co-polymer.

Figure 2. In vitro hydrolytic degradation of RAM:SA50:50 polymer monitored by wet and dry weight lossstudy. A decrease in dry and wet weight can be observedwith time because of simultaneous increase in porosity ofthe device.



ence the mechanism and rate of degradation; type ofmonomers and their composition are most importantattributes.23 It have been reported that as the fractionof fatty acid increases in the polyanhydrides thepolymer no longer remains only-surface erodingpolymer.3,15 The weight loss study supports the roleof surface erosion in degradation of the synthesizedpolymer. A supplementary espouse to the earlierfact is decrease in size of the device in a manner

resembling surface degradation. Hence, it can bepredicted that the polymers with high proportion ofSA are major surface degrading polymer and as wego along increasing ratio of RAM, the surface erod-ing property declines.

‘‘Anhydride loss’’ is the diminution of the anhy-dride and intensification of the acid peak in FTIRduring degradation of the polymer.24 This particularphenomenon occurs due to generation of carboxylic

Figure 3. Scanning electron micrographs (SEM) of polymer RAM:SA 50:50 duringdegradation. The polymer demonstratesincrease in surface porosity with time. (magnification of 1.25 KX).



acid group as a result of hydrolysis of anhydridebonds. It was found that the polymer looses around50% of anhydride bond within 24–48 h (Fig. 4) andthis is in concordance with the previously reporteddegradation studies with similar kind of polymer.4

Rapid anhydride loss does not lead to fast erosion ofthe polymer because anhydride cleavage is notdirectly linked to existence of the matrix, whichremains as such. A strong correlation between anhy-dride loss and drug release could not be made but itcan serve as a characteristic property of the polymer.

In vitro drug release studies

In vitro release studies were carried out with devi-ces of all the polymers synthesized, containing 5%MTX, in phosphate buffer (pH 7.4) at 378C. Drugrelease was evaluated for 240 h as shown in Figure 5.

It can be observed that release rate increases as afunction of RAM content. RAM:SA 10:90 polymerreleases 63% of the drug in 10 days and the amountof drug released has increased proportionally toRAM. Polymers with 70:30 and 90:10 RAM:SAreleases the drug relatively at rapid rate, which canbe explained by the exposure of larger surface areato the release media as these polymer samples cannot be cast into the definite shape because of their li-quidity, hence 200 lL of polymer incorporated withdrug was injected in the release media. Polymershaving RAM content in range of 30–60% have notshown any major disparity in their release profiles(Fig. 5). The release of incorporated substance fromany polymeric device can be elucidated by three im-portant aspects, which include: polymer erosion;polymer swelling; and diffusion process. Release ofthe drug in a system is mainly controlled by the fac-tor which takes up velocity with time. Transforma-tion from one form of release pattern to other is alsopossible, for example, system which initially swellsor erodes rapidly, both of which increase the poros-ity can later turn out to be a diffusion controlled sys-tem. MTX release from the device in this study canalso be explained in the similar manner based on thecorrelation coefficients values, calculated by fittingthe release profiles in four different models.25 Poly-mers with higher ratio of SA (70 and 90%) haveshown nearly zero order release, which is the charac-teristic of polyanhydrides. On the other hand, poly-mer with the increasing proportion of RAM show agradual drift in the release profile away from zeroorder except in case of 90:10 RAM:SA, where the R2

is higher than 70:30 RAM:SA polymer. This may be

Figure 4. Anhydride loss occurred during degradation ofRAM:SA 50:50. Hydrolysis was rapid initially and becomegradual as the time progressed.

Figure 5. Release profiles of methotrexate from polymers with different ratios of RAM:SA. The study was conducted inphosphate buffer (0.1M, 7.4 pH) at 378C and 100 RPM. As the proportion RAM increased the release rate found to befaster, loading was 5% w/w in all cases. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]



due to the shape of the device which polymer takesafter injection into the release media.26

The synthesized polymers, which are almost semi-solid, can be considered as similar kind of system asproposed in Higuchi model.25 This model entailsthat, after some time the total drug concentrationgradient in a delivery system would be expected toshow a sharp discontinuity no drug dissolution canoccur until the concentration drops from initial drugconcentration to below the matrix solubility. At thesame time Hixson–Crowell25 model implies linearcorrelation of cubic root of the unreleased fraction ofdrug versus time. The model is applicable to the sys-tems where geometrical shape of the device dimin-ishes proportionally over time and this is particu-larly true for the devices cast with the drug loadedsynthesized polymers (Fig. 6). Moreover, the drugrelease from all the polymers have shown phasesand a slow depletion of the drug from layers, sur-face erosion of that layer so that next layer againleads to high concentration gradient, which furtherreleases the drug and the process continues.

Slow release from 10:90 RAM:SA polymer can beexplained by the crystallinity of SA monomer. SA hasreported crystallinity of around 66%, which in turnfavor the integrity of the matrix device and thusresulted in slow release rate. Contrary to this, anincrease in RA content increases the fluidity of thepolymer with a large surface area resulting higherdrug release rates. It is the proven fact that the crystal-line polymer degrades slowly and hence results inslow drug release.19,27,28 Increasing the RAM contentdecreased the melting point and crystallinity of thepolymers, which resulted in comparatively fasterrelease from other polymers.

Effect of drug loading on release

An important factor in the design of a polymericcontrolled release system is the percentage of drug

loading. Drug loading determines the geometryand the topology of the channels, which dictate therelease of the drug molecules at the surface, allowthe exposure of inner drug to media.29 No effect ondrug release was found with increase in MTX load-ing from 5, 10, and upto 20% (Fig. 7), which is anattribute for surface eroding polymer. Polymerswith properties of bulk degradation and swellingare diffusion controlled systems and affected bydrug loading.29 While in case of surface erodingpolymers it is expected that the rate of drug releaseshould not be affected by loading provided thatsink conditions are maintained and loading is notvery high. In the later case very high loading canlead to a device in which amount of release con-trolling polymer becomes insignificant to controlthe release of drug.

Figure 6. Reduction in the size of drug loaded polymer (RAM:SA, 50:50) device with time. It indicates the surface erosionbut at the same time it can also be observed that some initial points show drug depletion from the surface. [Color figurecan be viewed in the online issue, which is available at www.interscience.wiley.com.]

Figure 7. Effect of drug loading on cumulative percentrelease of MTX from polymer RAM:SA 50:50. Experimentwas performed in triplicate. Surface eroding polymerslike polyanhydride are believed to be not affected byloading.



Effect of drug solubility on release: Methotrexateversus 5-fluorouracil

Release from polymer matrix depends on thephysicochemical properties of the incorporated drug,and it frequently plays a significant role in the rateof drug release from bioerodible systems.30–32 Solu-bility of the drug is one of the critical factors, whichdecide release rate. 5-FU has water solubility of96 mM and molecular weight of 130.08, while MTXhave molecular weight of 470.48 and shows pH de-pendent solubility, ranging from 0.9 mM at pH 5 to20 mM at pH 7.33 In a comparative study of MTXand 5-FU release, it was found that release of 5-FUwas relatively faster (Fig. 8) and the major factorwhich leads to this behavior is high solubility of5-FU in comparison to MTX. Another contributingfactor for higher release rate can be the low molecu-lar weight of 5-FU, which makes it easy for themolecules to come out of the pores. When release of5-FU was studied using various dissolution models(zero order, first order, Higuchi and Hixson–Cro-well), it was revealed that release pattern was princi-pally controlled by erosion together with petite con-tribution of diffusion.

Erosion of biodegradable polyanhydride device isnot simply a function of polyanhydride chemistrybut also involves water uptake, diffusion, and disso-lution of the degradation products and hence is afunction of the nature of the compounds incorpo-rated into the device. It is reported that water-solu-ble compounds will be released rapidly creatingpores and channels in the devices, which will bepenetrated by water, resulting in faster erosion ofthe device compared with placebo devices.30 While

poorly water soluble compounds retard erosion.From product development standpoint, effect of thedrug’s physicochemical characteristics on device ero-sion should be considered to determine its in vivolife time and disappearance.

In vivo evaluation of polymers for biocompatibility

Animals were injected with 500 lL of polymers inrat back at neck region by gently warming RAM:SA50:50 and 70:30 polymers at 45 and 408C, respec-tively. The polymer became injectable liquid. Thecontrol group was injected with same amount ofnormal saline to make a comparison keeping similarenvironment. When animals were sacrificed atscheduled time intervals the polymers were found insolid state at the injection site. Size of these in situformed implants was found to be reduced with timeindicating the in vivo degradation of the polymer.All the animals were healthy throughout the experi-ment period, of 22 days, as judged from the bodyweight of the treated and control rats during theexperiment. To compare the change in body weightof control and treated group, student t-test wasapplied and no significant difference (p > 0.05)between the groups was found. Animals were sacri-ficed for histopathological study at five differenttime points (1, 3, 7, 12, and 22 days) and no grosspathological findings were noted at the injection siteexcept trauma due to injection in three animals (Twofrom group with 50:50 RAM:SA polymer and onefrom control group). The tissue nearby injection sitewas processed for microscopy and a thin section of5 lm was stained with hematoxylin and eosin dyefor histopathological examination. Each section wasgiven a grade for inflammation on an arbitrary scaleof 0–4 as explained earlier in experimental part. Theaverage scores at each time period for all groups aregiven in Table II. It was observed that there was nomajor acute inflammation due to polymer after1 day of injection but subacute host response sets inagainst the foreign material. Inflammation increases

Figure 8. Effect of drug solubility on release profile. Dur-ing the study period 5-FU (–^–) release was faster thanMTX (–u–) because of comparatively high solubility of 5-FU. [Color figure can be viewed in the online issue, whichis available at www.interscience.wiley.com.]

TABLE IIAverage Score of Inflammation for Accessing

Biocompatibility of the Polymers

Day ControlRAM:SA50:50

RAM:SA70:30

1 0 3.5 2.53 1 2.5 4.07 0 3.0 4.012 2 4.0 2.522 1 1.5 1.5

The score is the average of inflammation grade assignedto an individual tissue on arbitrary scale of 0–4.



Figure 9. Representative histopathological tissue samples of animals. Injected with normal saline, polymer RAM:SA 50:50and 70:30, stained with hematoxylin and eosin dyes at 203. Signs of high inflammation in a few of the tissue samples(RAM:SA 70:30, 3 and 12 days) but usually there was an indication of time-related healing process P(RAM:SA) 50:50 andP(RAM:SA) 70:30 for 22 days. Polymer degradation can also be noticed, as crystals of SA detected (showing birefringence)in the 22 days tissue sections (shown by arrows). [Color figure can be viewed in the online issue, which is available atwww.interscience.wiley.com.]



from 3, 7, to 12 days and some of samples haveshown grade 4 response, for example, RAM:SA 70:30on 3rd and 12th day (Fig. 9). It can be due to lowmelting point of the polymer, which leads to largercontact area between polymer samples and SC tis-sue. Although, there were some signs of high inflam-mation in a few of the tissue samples but usuallythere was an indication of time-related healing pro-cess (i.e., time-related reduction in incidence and se-verity of necrosis and acute to subacute inflamma-tory reaction, associated with increased fibroplasia)(Fig. 9). Control group has also shown mild inflam-mation due to trauma of injection. On day 22, theinflammation decreased and the tissue shown signsof healing. Variations in the animal response havealso been observed, as same group of animals at thesame time points have shown different degree ofinflammation. One more interesting finding from thesections of 22 days was presence of SA crystalswhich appeared in both the polymer. It confirms thein vivo degradation, the crystals have shown birefrin-gence in polarized light.

CONCLUSION

RAM and SA were copolymerized in variousratios by a one pot condensation method. Polymerswere low molecular weight and have showndecrease in the melting temperature with increase inRAM content. Polymer degraded majorly by surfaceerosion but at the same time presence of petite bulkerosion also appeared. MTX release pattern for highSA content of polymer was found to be nearly zeroorder, on the other hand high RAM ratio lead tohigher release rate, which follow mix Higuchi andHixson–Crowell model. Different level of MTX load-ing (5, 10, and 20%) did not affect the release to amajor extent because of surface eroding property ofthe polymer. Degradation of RAM:SA 50:50 polymeris representative however all mentioned polymersare useful depending on the requirement of the du-ration of drug release. RAM:SA 70:30 would be thepolymer of choice due to its better injectability if lon-ger release is not required. Solubility of the incorpo-rated drug also affect the release profile as the watersoluble 5-FU release faster than MTX. The polymerwas evaluated for biocompatibility and initialinflammation was observed from the histopathologyof the tissue near the injection site, which has showntime-related healing. Hence these polymers can beconsidered as a potential drug carrier for localizeddelivery of anticancer drugs.

JPJ and SM are thankful to NIPER for the fellowship tocarryout this work. Authors are also thankful to Dr. A. C.Dutta for helpful discussion on histopathological studies.

References

1. Jain JP, Modi S, Domb AJ, Kumar N. Role of polyanhydrides

as localized drug carriers. J Control Release 2005;103:541–563.2. Rosen HB, Chang J, Wnek GE, Linhardt RJ, Langer R. Bioer-

odible polyanhydrides for controlled drug delivery. Biomate-rials 1983;4:131–133.

3. Teomim D, Domb AJ. Fatty acid terminated polyanhydrides.

J Polym Sci Part A: Polym Chem 1999;37:3337–3344.4. Teomim D, Nyska A, Domb AJ. Ricinoleic acid-based biopol-

ymers. J Biomed Mater Res 1999;45:258–267.5. Deng JS, Meisters M, Li L, Setesak J, Claycomb L, Tian Y, Ste-

phens D, Widman M. The development of an injection-mold-

ing process for a polyanhydride implant containing gentami-

cin sulfate. PDA J Pharm Sci Technol 2002;56:65–77.6. Kumar N, Krishnan M, Azzam T, Magora A, Ravikumar

MNV, Flanagan DR, Domb AJ. Analysis of fatty acid anhy-

drides and polyanhydrides. Anal Chim Acta 2002;465:257–

272.7. Shen E, Pizsczek R, Dziadul B, Narasimhan B. Microphase

separation in bioerodible copolymers for drug delivery. Bio-materials 2001;22:201–210.

8. Shikanov A, Vaisman B, Krasko MY, Nyska A, Domb AJ Pol-y(sebacic acid-co-ricinoleic acid) biodegradable carrier forpaclitaxel: In vitro release and in vivo toxicity. J Biomed MaterRes A 2004;69:47–54.

9. Franson R, Regelson W, Ottenbrite R. Cytoprotective fattymoiety compounds. Patent PCT WO 91/03512, 1991 March.

10. Najib N, Suleiman M. Kinetics of drug release from ethylcel-

lulose solid dispersions. Drug Dev Ind Pharm 1985;11:2169–

2181.11. Higuchi T. Rate of release of medicaments from ointment

bases containing drugs in suspension. J Pharm Sci 1961;50:874–875.

12. Higuchi T. Mechanism of sustained-action medication. Theo-

retical analysis of rate of release of solid drugs dispersed in

solid matrices. J Pharm Sci 1963;52:1145–1149.13. Hixson AW, Crowell JH. Dependence of reaction velocity

upon surface and agitation. Ind Eng Chem 1931;23:923–931.14. Bruneton J. Lipids: Vegetable oils. In: Bruneton J, editor.

Pharmacognosy Phytochemistry of Medicinal Plants. New

Jersey: Lavoisier Publishing; 2001. p 137–166.15. Teomim D, Domb AJ. Nonlinear fatty acid terminated poly-

anhydrides. Biomacromolecules 2001;2:37–44.16. Shieh L, Tamada J, Tabata Y, Domb A, Langer R. Drug

release from a new family of biodegradable polyanhydrides.

J Control Release 1994;29:73–82.17. Laurencin CT, Gerhart T, Witschger P, Satcher R, Domb A,

Rosenberg AE, Hanff P, Edsberg L, Hayes W, Langer R. Bio-

erodible polyanhydrides for antibiotic drug delivery: In vivo

osteomyelitis treatment in a rat model system. J Orthop Res

1993;11:256–262.18. Hill K. Fats and oils as oleochemical raw materials. Pure

Appl Chem 2000;72:1255–1264.19. Mathiowitz E, Ron E, Mathiowitz G, Amato C, Langer R.

Morphological characterization of bioerodible polymers. I.

Crystallinity of polyanhydride copolymers. Macromolecules

1990;23:3212–3218.20. Maglio G, Migliozzi A, Palumbo R. Thermal properties of di-

and triblock copolymers of poly(L-lactide) with poly(oxyethy-

lene) or poly(e-caprolactone). Polymer 2003;44:369–375.21. Masamoto J, Yajima K, Sakurai S, Aida S, Ueda M, Nomura

S. Microphase separation in polyoxymethylene end-capped

with a long-chain alkyl. Polymer 2000;41:7283–7287.22. Zhao Y-F, Fan X, Chen X, Wan X, Zhou Q-F. Synthesis and

characterization of diblock copolymers based on crystallizable

poly(e-caprolactone) and mesogen-jacketed liquid crystalline

polymer block. Polymer 2005;46:5396–5405.



23. Gopferich A, Tessmar J. Polyanhydride degradation and ero-sion. Adv Drug Deliv Rev 2002;54:911–931.

24. Santos CA, Freedman BD, Leach KJ, Press DL, Scarpulla M,Mathiowitz E. Poly(fumaric-co-sebacic anhydride). A degra-dation study as evaluated by FTIR, DSC, GPC and X-ray dif-fraction. J Control Release 1999;60:11–22.

25. Costa P, Lobo JMS. Modeling and comparision of dissolutionprofiles. Eur J Pharm Sci 2001;13:123–133.

26. Akbari H, D’Emanuele A, Attwood D. Effect of geometry onthe erosion characteristics of polyanhydride matrices. Int JPharm 1998;160:83–89.

27. Deng JS, Li L, Tian Y, Meisters M, Chang HC, Stephens D,Chen S, Robinson D. The relationship between structures andin vitro properties of a polyanhydride implant containinggentamicin sulfate. Pharm Dev Technol 2001;6:541–549.

28. Chan CK, Chu IM. Crystalline and dynamic mechanicalbehaviors of synthesized poly(sebacic anhydride-co-ethyleneglycol). Biomaterials 2003;24:47–54.

29. El-Arini SK, Leuenberger H. Modelling of drug release frompolymer matrices: Effect of drug loading. Int J Pharm 1995;121:141–148.

30. Park ES, Maniar M, Shah J. Effects of model compounds withvarying physicochemical properties on erosion of polyanhy-dride devices. J Control Release 1996;40:111–121.

31. Park ES, Maniar M, Shah J. Influence of physicochemicalproperties of model compounds on their release from bio-degradable polyanhydride devices. J Control Release1997;48:67–78.

32. Park ES, Maniar M, Shah JC. Biodegradable polyanhydridedevices of cefazolin sodium, bupivacaine, and taxol forlocal drug delivery: Preparation, and kinetics and mecha-nism of in vitro release. J Control Release 1998;52:179–189.

33. Rubino FM. Seperation methods for methotexate, its struc-tural anlogues and metabolites. J Chrom B 2001;764:217–254.



Hydroxy fatty acid based polyanhydride as drug delivery system: Synthesis, characterization, in vitro degradation, drug release, and biocompatibility

Documents