j ourna l homepage: www.e lsev ie r.com/ locate / jconre l
Drug release and swelling kinetics of directly compressed glipizide sustained-releasematrices: Establishment of level A IVIVC
Jolly M. Sankalia ⁎, Mayur G. Sankalia, Rajashree C. MashruCentre of Relevance and Excellence in Novel Drug Delivery Systems, Pharmacy Department, The M. S. University of Baroda, G. H. Patel Building, Donor's Plaza, Vadodara,Gujarat 390 002, India
as to examine a level A in vitro–in vivo correlation (IVIVC) for glipizide hydrophilicsustained-release matrices, with an acceptable internal predictability, in the presence of a range of formulation/manufacturing changes. The effect of polymeric blends of ethylcellulose, microcrystalline cellulose, hydro-xypropylmethylcellulose, xanthangum,guargum,Starch1500, and lactose on invitro releaseprofileswas studiedandfitted to various release kineticsmodels.Water uptake kineticswith scanningelectronmicroscopy (SEM)wascarried out to support the drug releasemechanism. An IVIVCwas established by comparing the pharmacokineticparameters of optimized (M-24) and marketed (Glytop-2.5 SR) formulations after single oral dose studies onwhite albino rabbits. The matrix M-19 (xanthan:MCC PH301 at 70:40) and M-24 (xanthan:HPMC K4M:Starch1500 at 70:25:15) showed the glipizide release within the predetermined constraints at all time points withKorsmeyer–Peppas' and zero-order release mechanism, respectively. Kopcha model revealed that the xanthangum is the major excipient responsible for the diffusional release profile and was further supported by SEM andswelling studies. A significant level A IVIVC with acceptable limits of prediction errors (below 15%) enables thepredictionof in vivo performance from their in vitro release profile. Itwas concluded that proper selection of rate-controlling polymers with release rate modifier excipients will determine overall release profile, duration andmechanism from directly compressed matrices.
In the last few years, diabetes mellitus has reached epidemicproportion and is now becoming cause of premature mortality andmorbidity [1]. People with type II diabetes mellitus are characterizedby a pancreatic β-cell dysfunction, resistance to insulin and a relative,as opposed to absolute, insulin deficiency [2]. Glipizide, a secondgeneration sulfonylurea, is used for patients with type II diabetes whohave failed diet and exercise therapy and it appears to be the mosteffective insulin secretagogue both in first phase insulin secretion andin sustained stimulatory response during long term administration. Ablood glucose concentration decreases in about 30 min of ingestion,providing peak concentrations within 1–3 h after a single oral dose ofglipizide with an elimination half-life of about 2–4 h. Such a rapidlyabsorbed drugs having faster elimination rate with short half-lifemake it suitable candidate to be formulated for the sustained delivery.
Generally, primary objectives of controlled drug delivery are toensure safety and to improve efficacy of drugs as well as patientcompliance, which can be achieved by better control of plasma drug
levels and less frequent dosing [3]. Themost convenientway to achievecontrolled release of active agent involves physical blending of drugwith polymer matrix, followed by direct compression, compressionmolding, injectionmolding, extrusion, or solvent castingwhich resultseither in monolithic device or in swellable hydrogel matrix [4–8]. Forany controlled-release dosage form it is very important to useminimum number of excipients with minimum processing steps inorder to reduce the tablet-to-tablet and batch-to-batch variations,hence direct compression is the most suitable and easily up-scalabletechnique [9]. When taken as an aggregate, directly compressedhydrophilicmatrices are the demandof today's fast going erawith botha scientific and economic appeal. Since the cost of synthesizing a newpolymeric substance and testing for its safety is enormous [10], a newfocus has been directed towards investigating the use of polymerblends of pharmaceutically approved polymeric materials as matrixexcipients to retard drug release. In vitro–in vivo correlations (IVIVCs)can decrease regulatory burden by decreasing the number ofbiostudies required in support of a drug product. Further, IVIVC canalso allow setting of more meaningful dissolution specifications. Boththe regulatory agencies and industry sponsors have understood thisvalue of IVIVCs. Therefore, the activity in the area of IVIVC for oralextended-release dosage forms has increased in the last 5 years [11].Hence, the investigations were aimed to study different formulationvariables and design the directly compressed glipizide sustained-
50 J.M. Sankalia et al. / Journal of Controlled Release 129 (2008) 49–58
release oral hydrophilic matrices and to establish a level A IVIVC withinternal predictability.
2. Materials and methods
2.1. Materials
Glipizide (99.79% pure) was kindly supplied by Apostle Remedies(Vadodara, India). Gliclazide (purity of 99.69%)was obtained as gift fromRelax Pharmaceuticals (Vadodara, India). Samples of Glytop 10 SR®
(10 mg glipizide per tablet) and Glytop 2.5 SR® (2.5 mg glipizide pertablet, RPG Life Sciences, India)were obtained from retail pharmacy andused as reference product. Ethylcellulose (EC, Ethocel® standard 7 FPpremium), hydroxypropylmethyl cellulose (HPMC, Methocel® K100LV,K4M, K15M, K100M grade) and Starch 1500® were kindly gifted by theColorcon® Asia Pacific (Mumbai, India). Microcrystalline cellulose (MCC,Avicel® PH 301), was gifted by the FMC Biopolymers® (Ireland). Xanthangum USP XXVIII (Jungbunzlaver, Austria), lactose spray-dried (Flowlac100®, Meggle, Germany), guar gum (Amit Cellulose Products, Mumbai,India), colloidal silicon dioxide (Aerosil SD 200®, Degussa AG, Germany),magnesium stearate (Mallinckrodt, USA, used as a lubricant in tableting)were obtained as gift samples from Relax Pharmaceuticals (Vadodara,India). Deionized double-distilledwaterwas used throughout the study.All other chemicals and solvents were of analytical grade and usedwithout further purification.
2.2. Preparation and characterization of matrix tablets
Drug–excipient interactionwas investigated by differential scanningcalorimetry (DSC-60, Shimadzu, Japan) from 30–400 °C at a heating rate
Table 1The preliminary experimental plan used to investigate the influence of MCC, HPMC, starch agum based matrices along with response variables used in the study with constraints
Q(2) = percent dissolved in 2 hQ(4) = percent dissolved in 4 hQ(6) = percent dissolved in 6 hQ(8) = percent dissolved in 8 hQ(10) = percent dissolved in 10 hQ(12) = percent dissolved in 12 h
Values in table show weight of respective excipient in mg. The amount of glipizide, silicon dtable). Matrix composition in bold figures resulted in drug release within the constraints pr
of 10 °C/minunder constantpurgingof drynitrogen at30ml/min. Beforecompression, the powder mixtures were checked for Hausner ratio,Carr's compressibility index, and angle of repose. Preliminary experi-ments were carried out by preparing different matrix tablets by directcompression of several homogenous blends of different polymers andexcipients in different ratios in order to achieve desired drug releaseprofile (Table 1). The ability of the EC, guar gum and xanthan gum as adrug release retardant in the direct compression formulations and theirperformance mechanism were evaluated. An overview of matrix codeand unit formula of all matrices evaluated during the study is shown inTable 1 (M-1 to M-25) and divided in sub-groups depending on theformulation composition. Briefly, weighed quantity of drug wasphysically mixed with all the auxiliary excipients by geometric additionusing a glass mortar and pestle for about 10 min. Then magnesiumstearate and silicon dioxide were added as the glidants/lubricant andthoroughly blended for 2 min and were sieved through ASTM sieve #24sieve (0.710mm). The homogeneous powdermixture for a singlematrixwas weighed, fed manually into the die of an eight station automaticrotary tablet machine (Modern Engineering Works, New Delhi, India)equipped with flat faced die-punch set of 6.81 mm diameter, andcompressed to a target weight (as per the composition of the matrix)and an average hardness of 6–7 kg/cm2 for all the tablets. The obtainedmatriceswere subjected to various physico-chemical investigations like,appearance, weight variation, thickness, hardness, drug content, and invitro drug release. For in vivo study, 50 matrix tablets weighing 21 mg(contains 1.87 mg glipizide per tablet) were prepared separately using3 mm Ø×4 hole die and 4 flat tip upper–lower punch set (Jyoti MechIndustries, Navsari, Gujarat, India) on above mentioned tablet machine.The composition of the small tablet was kept proportionally similar tothat of the optimized batch.
nd lactose on the release profile of glipizide from ethyl cellulose, guar gum and xanthan
ioxide and magnesium stearate was fixed at 10, 1 and 2 mg, respectively (not shown inedetermined.
51J.M. Sankalia et al. / Journal of Controlled Release 129 (2008) 49–58
2.3. Estimation of glipizide
Glipizide content from powder mixture, matrices and in vitrodissolution study samples were measured by spectrophotometricShimadzu UV-1601, Shimadzu Corporation, Kyoto, Japan) methoddeveloped inphosphate buffer pH6.8 (IP 1996) over an analytical rangeof 1–50 μg ml−1 at 276 nm. An HPLC (Shimadzu, Kyoto, Japan) methodhas been developed and validated for quantitation of glipizide(gliclazide used as internal standard) from rabbit plasma samples.Analysiswas performedonPhenomenex C18 columnat 1.0ml/minflowratewith 70:30 v/vmixture of acetonitrile: 10mMKH2PO4 buffer pH 3as mobile phase (MP) at 225 nm.
2.4. In vitro studies
2.4.1. Drug release studiesThe USPmonographs for extended-release dosage forms specify the
percent of drug released after more than one time point. Therefore, thepercent of drug release after 2, 4, 6, 8, 10 and 12 h (Qt) with constraintwas selected as the response variables (Table 1). The dissolution of thedeveloped matrix formulations (n=6) was carried out in USP type II(paddle) apparatus using phosphate buffer pH 6.8 (IP 1996) at 100 rpmand 37±0.5 °C up to 12 h (TDT-60T, Electrolab, Mumbai, India). Twomilliliter dissolutionmediumwaswithdrawnat 0, 0.25, 0.5, 0.75,1,1.5, 2,3, 4, 5, 6, 7, 8, 9,10,11, and 12 h,filtered and analyzed spectroscopically. Amean dissolution time (MDT) for 50% or 80% release of drug werecalculated using the following expression (Eq. (1)):
MDT ¼Xnj¼1
t̂jDMj=Xnj¼1
DMj ð1Þ
where j is the sample number, n is the number of dissolution sampletimes, t̂j is the time at midpoint between tj and tj−1 (easily calculatedwith the expression (tj+tj−1) /2) and ΔMj is the additional amount ofdrug dissolved between tj and tj−1.
2.4.2. Curve fitting and release kineticsIn order to propose the possible release mechanism, the release
pattern was evaluated to check the goodness of fit for zero-orderrelease kinetic, Higuchi's square root of time equation [12], Kors-meyer–Peppas' power law equation [13,14] and Hixson–Crowell'scube root of time equation [15]. The goodness of fit was evaluated by r(correlation coefficient) values. In order to understand the releasemechanism, the release data of the optimized batch (M-6) was fittedto empirical equations proposed by Kopcha [16] (Eq. (2)),
M ¼ At1=2 þ Bt: ð2Þ
In the above equations, M (≤70%) is the percentage of drug re-leased at time t, while A and B are, respectively, diffusion and erosionterms. According to this equation, if diffusion and erosion ratio, A/B=1,then the release mechanism includes both diffusion and erosionequally. If A/BN1, then diffusion prevails, while for A/Bb1, erosionpredominates [17]. The mean release profile (n=12) of optimizedbatch was compared with that of the commercial formulation (Glytop10 SR®) using the model independent pair-wise approach, thesimilarity factor (f2) [18].
described method [19]. Briefly, initial diameter, height and weight ofindividualmatrices (S0)weremeasured andwere placed in a dissolutionmedium (phosphate buffer pH 6.8) at 37±0.5 °C. Swollen/hydratedtablets were withdrawn from the medium, extra buffer present on thematrix surface was gently wiped with the soft tissue, and individualdiameter, height andweight (S1) weremeasured at predetermined time
intervals. Percent of the radial (diameter) and axial (height) swelling oftablet and percent water uptake was calculated according to the fol-lowing formula (Eq. (3)):
where S1 and S0 are the diameter or height of swollen and dried tablets,respectively. The percentage swelling of the original tablet was cal-culated and plotted vs. time to reflect the rate of hydration.
2.4.4. Scanning electron microscopy (SEM)The surface morphology of the optimized batch before and after in
vitro dissolution at different time intervals was analyzed by SEM. Thetablet wasmounted on brass stubs using carbonpaste. SEM imagesweretaken using a scanning electron microscope (JSM-5610LV; Jeol Ltd.,Tokyo, Japan) at the required magnification at room temperature. Aworkingdistanceof 39mmwasmaintained, and the accelerationvoltageused was 10 kV, with the secondary electron image (SEI) as the detector.
2.4.5. Stability studyThe optimized glipizide formulations were strip packed (Al–Al
strip, 0.04mm) and subjected to accelerated stability studies as per ICHguidelines (40 °C±2 °C/75% RH±5% RH). The samples werewithdrawnperiodically (0, 15, 30, 60, 90, and 180 days) and evaluated for thedifferent physico-chemical parameters viz. appearance, weight varia-tion, thickness, hardness, drug content, and in vitro release studies.
2.5. In vivo studies
2.5.1. Animal housing and handlingThreewhite albino rabbits per groups, weighing 3.8–4.0 kg,were used
in the study in accordance with a protocol approved by the Institutionalethical committee at theM. S. University of Baroda at Vadodara, India. Theexperiments were conducted as per CPCSEA (Committee for Prevention,Control and Supervision of Experimental Animals) guidelines. All rabbitswere housed individually in cages under environmentally controlledconditions (23±2 °C; 55±5% relative humidity, 12 h light/dark cycle). Allrabbits were housed with free access to food and water; except for thefinal 24 h before experimentation.
2.5.2. Pharmacokinetic analysisA single-dose pharmacokinetic study was conducted onwhite albino
rabbits. The group I rabbits were administered glipizide oral solution(1.87 mg in 20 ml aqueous solution) by oral–gastric intubation forestimation of pharmacokinetics parameters. Group II and group III rabbitswere administered with marketed formulation Glytop® 2.5 SR and M-24(xanthan gum:HPMC K4M:Starch 1500 at 70:25:15), respectively.Following the administration, 2.0 ml of blood samples was collectedfrom themarginal ear vein at 0, 0.5,1, 2, 4, 6, 8,10,12,16, 20 and 24 h intoheparinized collection tubes. The blood was immediately centrifuged(1900 g) for 10 min at an ambient temperature. The supernatant plasmalayer was separated and stored at −20 °C until analyzed. The plasmasampleswere analyzed for glipizide concentrations by HPLCmethod. Thefirst order elimination rate constant (kel) was estimated by the least-square regression of plasma concentration–time data points of the curvesdescribing the terminal log-linear decaying phase. T1/2 was derived fromkel (T1/2=ln 2/kel, where ln is the natural logarithm). The area under theplasma concentration–time curve from zero to the last measurableplasma concentration at time t (AUC0− t) was calculated using the lineartrapezoidal rule. The area was extrapolated to infinity (AUC0−∞) byaddition of Ct /kel to AUC0− t, where Ct is the last detectable drugconcentration. The absorption rate constant (ka) was determined byresidual method[20]. The maximum observed glipizide concentration(Cmax) and the time at which Cmax was observed (Tmax) were reporteddirectly from the profile. The AUMC is the area under the plot of time vs.the product of time and concentration extrapolated to infinity and was
52 J.M. Sankalia et al. / Journal of Controlled Release 129 (2008) 49–58
calculated by trapezoidal rule. Volume of distribution (Vd) and totalclearance rate (TCR) were calculated using Eqs. (4) and (5), respectively.The mean residence time (MRT) was determined by AUMC divided byAUC. The clearance (Cl) was calculated as dose divided by AUC withextrapolation to infinity (AUC0−∞). All the pharmacokinetic parameterswere calculated using Microsoft® Office Excel 2003 (Microsoft Corpora-tion, USA) software application.
Vd ¼ D0G � AUMC
AUCð Þ2ð4Þ
TCR ¼ kel � Vd ¼ Vd � 0:693t1=2
: ð5Þ
The Wagner–Nelson method [21] was applied to deconvolute thepercentage of the glipizide absorbed using Eq. (6).
F tð Þ ¼ C tð Þ þ kelAUC 0�tð Þ ð6Þ
where, F(t) is the amount of drug absorbed. The fraction absorbed(FRA) was calculated using Eq. (7) and was plotted against the fractionof drug dissolved (FRD) at the same time and the linear regressionanalysis was used to examine the in vitro–in vivo relationship.
2.5.3. Establishment of in vitro–in vivo correlation (IVIVC)An IVIVC refers to the predictive mathematical model describing
the relationship between in vitro property of an extended-releasedosage form (usually the rate or extent of drug dissolution or release)and a relevant in vivo response, such as plasma drug concentration oramount of drug absorbed. A level A correlation is usually estimated bya two-stage procedure: deconvolution followed by comparison of thefraction of drug absorbed (FRA) to the fraction of drug dissolved (FRD).A correlation of this type represents a point-to-point relationship be-
Fig.1. (A–C) Effects of EC:HPMC ratio on the glipizide release profile (A) EC 7 FP:HPMC K4M, (7 FP:HPMC K4M at 50:50) tablet surface (D, E) and tablet cross-section (F) after dissolution
tween in vitro dissolution and the in vivo input rate [22]. Accordingly,the deconvolution procedure was done on optimized formulation usingmean glipizide plasma concentration vs. time profile using the methodof Wagner–Nelson. The data of FRA (obtained by deconvolution) andFRD for each formulationwere then plotted to develop an IVIVC model.The linear, quadratic, and cubic models (Eq. (8)) bearing the followingforms were tested to obtain the best fit:
Y ¼ aþ bX; Y ¼ aþ bX þ cX2; Y ¼ aþ bX þ cX2 þ dX3 ð8Þ
where a, b, c, d represent regression parameters associated with eachfunction, Y is the FRA in vivo and X is the FRD in vitro. The coefficients ofdetermination (r2) for all developed models were determined. Thehighest r2 value equation was considered as the best fitting model.
2.5.4. Internal validation of IVIVCThe reliability of the developed IVIVC model was evaluated by
using the internal predictability as described in relevant regulatoryguidance [23]. One recommended approach is based on a convolutionprocedure that models the relationship between in vitro dissolutionand plasma concentration in a single step. Plasma concentrationspredicted from the model and those observed are compared directly[22]. For this method, a reference treatment using bolus input isdesirable. An aqueous oral solution can be the bolus input forsustained-release formulation by gastrointestinal route [24]. Thepredictability was tested as follows: The in vitro release data of eachformulation was transformed into its in vivo release data by using theIVIVC equation. The plasma concentration profiles were calculated bynumerical convolution (Eq. (9)) of the in vivo release data with thehelp of the weighting function.
c tð Þ ¼Z t
0cd t � uð Þ � rabs uð Þ � du ð9Þ
where cδ represents the concentration–time profile resulting froman instantaneous absorption of a unit amount of drug, which is
B) EC 7 FP:HPMC K15M, and (C) EC 7 FP:HPMC K100M. (D–F) SEM photograph of M-3 (ECat different time points.
Fig. 2. Effect of (A) EC 7 FP:MCC PH301:Starch 1500 ratio on glipizide release, (B) MCCPH301 and HPMC K4M on glipizide release from xanthan gum matrices, and (C) co-excipients on the glipizide release from xanthan gum matrices and Glytop 10 SR®.
53J.M. Sankalia et al. / Journal of Controlled Release 129 (2008) 49–58
typically from bolus intravenous injection or reference oral solutiondata, c(t) is the plasma concentration vs. time profiles of the testedformulations, rabs is the input rate of the oral solid dosage form and uis the variable of integration. The predicted plasma concentrationswere compared with the observed plasma data and the percentageprediction error (%PE, Eq. (10)) of the IVIVC model was calculatedwith regard to Cmax, AUC0 − t and AUC0 −∞. According to the relevantFDA regulatory guidance [22] the permissible %PE values for Cmax
and AUC should be less than ±15% for each product and less than 10%for the average.
% Prediction error PEð Þ ¼ observed value�predicted valueobserved value
� 100: ð10Þ
3. Results and discussion
3.1. Drug–excipient interaction studies
The DSC is carried out to understand the solid-state interaction intablets. The DSC thermograms of pure drug, individual excipients,drug–excipient physical mixture (proportion same as the tabletcomposition) were recorded. To evaluate the internal structuremodifications after compression of physicalmixtures into tablet, tabletpowder was also included in the study. Physical mixture showedsimple superposition of their separated component DSC curves.According to the DSC findings of the matrix tablet powder, no majorthermal event corresponding to chemical interaction was observed(and hence not shown). All these confirmed the suitability of allexcipients with glipizide to prepare controlled-release inert matrices.
3.2. Characterization of matrix tablets
The results indicated that all the tablets prepared in this studymeet the USP 29 requirements for weight variation tolerance [25].Drug content of all tablet formulations were found in the range of 98.0to 102.0%. The thicknesses, diameters, and hardness variation of theindividual tablet batches werewithin ±3 SD. Tablets of the same batchshowed consistent dissolution behavior with small standard devia-tions in the subsequent dissolution studies.
3.3. Estimation of glipizide
For the spectroscopic method, a regression equation was calcu-lated as y=0.0166·x+0.0093 (r2=0.9997), in which x is the concentra-tion (μg ml−1), and y is the absorbance. Experimental results haveshown that the presence of the tablet ingredients has no interferencewith the spectroscopic method. For and HPLC method, a linearcorrelation (y=59.8925 x+3.4306; r2N0.9944) was obtained betweenthe glipizide/IS peak areas ratios and glipizide concentrations over therange of 10–2500 ng/ml. The limit of detection and quantitation werefound to be 2.64 and 8.80 ng/ml, respectively. Typical chromatogramat the optimized condition gave sharp and symmetric peak withretention time of 3.5 and 4.7 min for glipizide and IS, respectively.Extraction efficiency of glipizide from rabbit plasma samples wassatisfactorily ranged from 89.34 to 95.73%, which confirm nointerference effects due to plasma components throughout therange studied.
3.4. In vitro studies
3.4.1. Drug release studiesFormulation compositions affect drug release rates due to
polymer–excipient interactions, drug–polymer interaction, as wellas modulating matrix swelling and erosion rates. Thus, to under-stand the functional contribution of each excipient to the different
polymers and natural polysaccharide based matrices (EC–HPMC,xanthan gum, guar gum), dissolution studies were performed andthe results are discussed below. Mean percent glipizide release (n=6)was plotted against time and relative standard deviation at each timepoint was below 10% which indicated the reproducibility of theresults.
3.4.2. Curve fitting and release kinetic studies‘In vitro’ release profile for matrix tables (M-1 to M-10) prepared
with EC and HPMC K4M, K15M, and K100M is shown graphically inFig. 1(A), (B), and (C), respectively. EC–HPMC tablets exhibited a much
54 J.M. Sankalia et al. / Journal of Controlled Release 129 (2008) 49–58
slower release during 0–4 h because of low water affinity of the EC[26,27]. As the relative composition of EC increased in the tablet, itretards the penetration of dissolution medium by providing morehydrophobic environment and thus cause delay in release of drugfrom the tablet. SEM photograph of EC–HPMC (M-3) tablet surface(after 3 and 6 h) and tablet cross-section after 6 h dissolution areshown in Fig. 1(D–F). It is obvious from the figure that there was noconsiderable change in the tablet surface even after 6 h. However, thematrix did not showmuch pores or channel and further confirms thatthe ECmatrix structure was intact. The physico-chemical properties ofthe EC do not seem to affect the release profile significantly after 4 h.The drug diffusion was expected to take place through a porousnetwork created by glipizide already dissolved within the matrixtogether with the initial voids in the matrix, filled by liquid medium[28]. After 4 h, glipizide release decreased with increase in HPMCcontent and viscosity/molecular weight. Viscosity of HPMC solutionsis the result of hydration of polymer chains, primarily through H-bonding of the oxygen atoms in the numerous ether linkages, causingthem to extend and form relatively open random coils. A given hy-drated random coil is further H-bonded to additional water molecules,entrapping water molecules within, and may be entangled with otherrandom coils. All of these factors contribute to larger effective size andincreased frictional resistance to drug diffusion. The drug release ratedecreased in the rank order HPMC K4MNK15MNK100M. Thus, HPMCwas found to be dominating excipient controlling the release rate ofglipizide in matrix tablets.
As shown in Fig. 2(A), both the co-excipients (MCC and Starch)exhibited substantial enhancement in the drug release rates from thetablets. The EC–MCC system can sustain the drug release up to 3 h,but as starch is added, the matrix resulted in complete drug releasewithin 0.7 h. The faster release rates observed with microcrystallinecellulose and starch might be due to its inherent disintegrant prop-erties [29–31]. It was not possible to obtain tablets for the for-mulation composition of M-14 to M-16. The tablet powder mixturesshowed poor carr's index value (below 5%) and it was difficult tocompress. Even at maximum compression force, the guar gum con-taining mixture resulted in hardness of 2 kg/cm2. Such a low hard-
Table 2An overview of the comparative characteristics of different drug release kinetic models, best fit
ness was not suitable for the handling, hence were discontinued forfurther study.
Xanthan exhibits pseudoplasticity (shear-reversible property)in aqueous solutions, which can be explained on the basis of itshelical structure. Viscosity of xanthan gum increases due to theunwinding of the ordered conformation such as helix into a ran-dom coil with a consequent increase in resultant shape and sizeof the molecules. The presence of anionic side chains on thexanthan gum molecules enhances hydration and makes xanthangum soluble in cold water. Nine xanthan gum based formulationswere screened to select those characterized by a minimal bursteffect and a slow release of glipizide over 12 h (shown in Fig. 2B–C).At a fixed glipizide dose, the total content of xanthan gum shows adramatic change in their dissolution profile as shown in Fig. 2B (M-17 and M-18). At lower xanthan gum content, rapid swelling ofmatrices with less tight hydrogel structure resulted in higher initialdrug release followed by complete release within 9 h. Conversely atthe higher xanthan gum content, the initial drug release wasdiminished and drug diffuses slowly continuously for more than12 h. As the amount of xanthan gum in the matrix increased, therewould be a greater degree of hydration with simultaneous swellingwhich results in a lengthening of the drug diffusion pathway andreduction in drug release rate. It seems that, there is some thresh-old level for xanthan gum, within which the slight difference in itsconcentration can result in statistically significant different drugrelease profiles.
Partial replacement of xanthan gum by MCC PH301 caused increasein drug release and (M-19 and M-20 as compared to M-18) andlinearization of the release profile (M-19). During dissolution, xanthangum absorbs water, swells, and becomes a hydrated gel. At the sametime, MCC PH301, having disintegration properties as discussed earlier,promoted the disintegration and erosion of thematrices, which resultedin higher drug release. Batch M-19 gave the glipizide release within thedefined constraints and was considered optimized for further study.Xanthan can produce much more viscous gel as compared to HPMCand similar findings were reported by other researchers also [32].When HPMC K4M was added to xanthan gum matrix (M-21), it
model, MDT50 andMDT80 for batchesM-1 toM-25 (Optimized batches are shown in bold)
7.14 0.96 −0.05 0.98 Zero order 2.43 4.073.24 0.98 −0.02 0.97 Zero order 5.87 –
55J.M. Sankalia et al. / Journal of Controlled Release 129 (2008) 49–58
swells considerably immediately and release the glipizide from thechannels created by dissolution medium penetration. However, astime elapses, xanthan and HPMC showed synergetic gelling abilityand produce such an extreme visco-elastic gel that only 75% ofglipizide released within 12 h. As shown in Fig. 2(C), the combi-nation of two different excipients with xanthan gum (M-22 to M-25)gave more linear release profiles as compared to above matricesof xanthan gum only or with single excipient (M-17 to M-21).This may be due to the alteration in the thickness, porosity and gelstructure of more complex peripheral hydrated layer through whichdrug diffuses out. Inclusion of MCC PH301, Starch 1500, and lactoseresulted in intermediate drug release rate as compared to M-17 andM-18. Hence, apart from providing acceptable flow and compression
Fig. 3. Results of (A) Kopcha model parameters and (B) swelling studies with images at diffformulation.
properties to the formulation, these auxiliary excipients contributedsignificantly in controlling the drug release without burst effector lag phase. Glipizide release of matrix M-24 containing xanthan,HPMC and starch fits within the constraint limits and was selectedas an optimized batch.
With EC–HPMC matrices, as the molecular weight/viscosity andcomposition of HPMC in matrix tablet increases, the release profilechanges from super case-II transport to Fickian diffusion mechanism(M-1 to M-10, Table 2). Tablets prepared with only xanthan gumfollows Higuchi's square root of time equation when in low con-centration, while follows Korsmeyer–Peppas' power law equation (M-17 and M-18). Matrices of xanthan with either MCC or HPMC releasesdrug by Korsmeyer–Peppas' equation at all studied combinations (M-
erent time interval for M-19 formulation. (C) and (D) represent same results for M-24
56 J.M. Sankalia et al. / Journal of Controlled Release 129 (2008) 49–58
19 to M-21). M-24 (xanthan–HPMC–starch) formulation comply withzero-order release mechanism as depicted in Table 2.
3.5. Characterization of optimized formulations
From the release profiles of EC–HPMC, EC–MCC, and xanthan gummatrices, M-19 andM-24were selected as the optimized formulationsas they fit all the release constrains. Both formulations were studiedfurther for release mechanism by Kopcha model [16], swelling study,SEM, and stability studies.
3.5.1. Release mechanism by Kopcha modelIt can be seen from Fig. 3(A) that matrix tablets M-19 (xanthan:
MCC PH301 at 70:40) showed predominance of erosion relative todiffusion for initial 2 h. This was contributed by the presence ofwater insoluble MCC. However, xanthan gum prepared strong geland the diffusion takes over erosion term after 2 h and was main-tained throughout the release profile. For the M-24 (xanthan:HPMCK4M:Starch 1500 at 70:25:15) (Fig. 3(C)) formulation showed matrixerosion as major release mechanism. However, the diffusion termincrease up to 5 h and become steady after that, it was less significantthan erosion.
3.5.2. Swelling studyMeasurement of swelling/hydration rates of different matrices
were carried out to gain insight into the observed phenomena of drugrelease with the rates of polymer hydration and to evaluate the extentof water penetration into the tablets. Results of water uptake, axialand radial expansion for M-19 are enumerated in Fig. 3(B) and thephotographs at different time intervals are also presented. It is evidentthat water uptake was continuously rising throughout the study. Thissupports the observation of Kopcha model that the diffusion waspredominant. The terminal water uptake and radial/axial swelling was
Fig. 4. Scanning electron micrograph of (A–C) M-19 (MCC PH301:xanthan gum at 40:70) anddifferent magnifications (pores created due to diffusion are shown by arrows).
less steep because the diffusional path length and distance to betraveled for dissolution media to reach the dry core increases withtime. Hence, the drug release profile fits into the Korsmeyer–Peppaspower law equation. Fig. 3(D) summarizes the results of water uptake,axial and radial expansion for M-24. Though water uptake was con-tinuously increasing with time similar to M-19, the radial and axialexpansion was almost constant after 6 h. This suggests the role oferosion tomaintain constant diffusional path length because of propersynchronization between erosion and diffusion. This ultimately re-leases the drug by zero order. Over hydration of the outer most gellayer and its erosion can be seen in the last photograph (24 h).
3.5.3. SEM studySurface structure of M-19 after 12 h dissolution was studied by
SEM analysis and shown in Fig. 4(A–C). From the above discussion, it isevident that the release was due to diffusion of drug and hence theopening of the channels can be seen in the form of pores on thesurfaces (shown by arrows). The higher magnification photographsshow viscous gel like rubbery structure. SEM images of M-24 surfacesafter 12 h dissolution are presented in Fig. 4(D–F). The micrographsfurther verify the hypothesis of surface erosion as discussed above inKopcha model and swelling study. None of the micrograph showedany pores or any viscous gel like structure but showed erodedsurfaces.
3.5.4. Stability studiesAccording to ICH guidelines (Q1E Step4), six month accelerated
stability study (40±2 °C/75±5% RH) for the optimized formulationsshowed negligible change over time for the parameters like appearance,weight variation, thickness, hardness, and drug content. The similarityfactor (f2) ranged from 84 to 96, with a 2 to 7% average difference. Shelf-life, calculatedby linearextrapolationby zero-orderkinetics,were foundto be 2.69 and 2.78 years for M-19 and M-24, respectively.
(D–F) M-24 (HPMC K4M:Starch 1500:xanthan gum at 25:15:70) after 12 h dissolution at
Fig. 5. (A) Mean plasma glipizide concentration vs. time profile after administration ofsingle dose of oral solution, Glytop® SR, and M-24 in white albino rabbits includingpredicted plasma concentrations for M-24. (B) Fraction dissolved (FRD) in vitro vs.fraction absorbed (FRA) in vivo for M-24.
57J.M. Sankalia et al. / Journal of Controlled Release 129 (2008) 49–58
3.6. In vivo studies
3.6.1. Pharmacokinetic analysisAfter glipizide oral solution (group I, n=3) administration, the drug
plasma concentrationsweremonitored for 24 h and shown in Fig. 5(A).The mean glipizide pharmacokinetic parameters (n=3) for oralsolution are summarized in Table 3. Similarly, Glytop® 2.5 SR and M-24 (xanthan gum:HPMC K4M:Starch 1500 at 70:25:15) were adminis-tered to group II and group III rabbits and the drug plasma con-centrations were monitored for 24 h as shown in Fig. 5(A). The resultswere calculated by non-compartmental analysis. The comparativeresults AUC0−∞ and Cmax with other pharmacokinetic parameters forglipizide oral solution, Glytop® 2.5 SR, and M-24 are enumerated inTable 3. The rapid decrease in glipizide concentration after adminis-tration of oral solution reflects the fast disposition and elimination ofthe drug. As can be seen form Table 3, Cmax, ka, and kel of the Glytop SR
Table 3Observed and predicted pharmacokinetic parameters for glipizide after administration of o
and M-24 were significantly lower than that of the oral solutionwhereas MRT and Tmax were significantly higher. This indicates thatthe developed formulation M-24 is a sustained-release formulationand its pharmacokinetic parameters are almost comparable to that ofthe marketed formulation.
3.6.2. Establishment of the IVIVCAn IVIVC can impart in vivomeaning to the in vitro dissolution test
and can be useful as surrogate for bioequivalence. Out of two opti-mized formulations, batch M-24 was subjected to in vivo study toestablish IVIVC and the samewas comparedwithmarketed sustained-release formulation Glytop® 2.5 SR for various pharmacokineticparameters (Table 3). A level A IVIVC between FRD and FRA for matrixtablets was investigated using linear–nonlinear regression anddepicted in Fig. 5(B). Good point-to-point relationship was observedfor M-24 matrices with regression coefficient of 0.9933 and slopeapproaching to unity, indicating a close correlation between the invitro release rate with their in vivo absorption.
3.6.3. Internal validation of IVIVCThe validity of correlations was assessed by determining how well
the IVIVC models could predict the rate and extent of glipizideabsorption as characterized by Cmax and AUC0−∞. The reliability of thedeveloped IVIVC model was evaluated by using the internal predict-ability procedure. According to US FDA guidelines, the permissiblepercent prediction error (PE) values for Cmax and AUC should be lessthan ±15% [23]. The in vitro release data of M-24was transformed intopredicted in vivo release data by numerical convolution using oralsolution data as the impulse function. The predicted plasma profile ofthe M-24 is shown in Fig. 5(A). The observed and predicted Cmax andAUC data along with %PE values are shown in Table 3 (below 15%). Theestablished level A IVIVC confirms the efficacy of proposed in vitromodel in simulating in vivo conditions. This kind of correlation is quiteimportant since it represents a point-to-point relationship between invitro dissolution and the in vivo input rate of the drug from the dosageform[33].
4. Conclusion
Overall, the findings of this study demonstrate that directcompression of drug release retardant xanthan gum with other rate-controlling excipients effectively controls glipizide release throughoutthe course of 12 h. Since the process of manufacturing-direct com-pression involves minimum number of unit operations, it is easily up-scalable without any sophisticated production facilities. Combiningxanthan gum with starch, MCC and HPMC at particular proportionsoffered promising sustained-release formulations (M-19 and M-24).The optimized formulations were characterized by the absence of aninitial burst and satisfactory sustained-release properties. Further-more, the in vitro release rate suggests that glipizide concentrationsmay be maintained at constant levels for 12 h. A level A IVIVC wasdeveloped and validated internally for optimized (M-24) sustained-release glipizide matrix. This IVIVC can be used to guide new productdevelopment, support SUPAC change, waive bioequivalence study andmore importantly, ensuring commercial product quality over the years.
ral solution, Glytop 2.5 SR, and M-24 to white albino rabbits (n=3)
58 J.M. Sankalia et al. / Journal of Controlled Release 129 (2008) 49–58
However, further investigations in human are required to prove theclinical usability of the experimental extended-release formulation.
References
[1] P. Zimmet, Addressing the insulin resistance syndrome: a role for the thiazolidi-nediones, Trends Cardiovasc. Med. 12 (2002) 354–362.
[2] A. Bhattacharyya, Treatment of type II diabetes mellitus, Hosp. Pharm. 8 (2001)10–16.
[3] V.H.K. Li, J.R. Robinson, V.H.L. Lee, Influence of drug properties and routes of drugadministration of the design of sustained and controlled release systems, in: J.R.Robinson, V.H.L. Lee (Eds.), Controlled Drug Delivery — Fundamentals andApplications, (Drugs and the Pharmaceutical Sciences), 2nd, vol. 29, MarcelDekker, Inc., New York, 1987, pp. 4–94.
[4] J. Heller, Use of polymers in controlled release of active agents, in: J.R. Robinson, V.H.L. Lee (Eds.), Controlled Drug Delivery— Fundamentals and Applications, (Drugsand the Pharmaceutical Sciences), 2nd, vol. 29, Marcel Dekker, Inc., New York,1987, pp. 179–212.
[5] H. Santos, F. Veiga, M.E. Pina, J.J. Sousa, Compaction, compression and drug releasecharacteristics of xanthan gum pellets of different compositions, Eur. J. Pharm. Sci.21 (2004) 271–281.
[6] S.M. Al-Saidan, Y.S.R. Krishnaiaha, V. Satyanarayanab, P. Bhaskarc, R.S. Karthikeyan,Pharmacokinetic evaluation of guar gum-based three-layer matrix tablets for oralcontrolled delivery of highly soluble metoprolol tartrate as a model drug, Eur. J.Pharm. Biopharm. 58 (2004) 697–703.
[7] G. Shlieout, G. Zessin, Investigation of ethyl cellulose as a matrix former and a newmethod to regard and evaluate the compaction data, Drug Dev. Ind. Pharm. 22(1996) 313–319.
[8] M.J. Tobyn, J.N. Staniforth, A.R. Baichwal, T.W. McCall, Prediction of physicalproperties of a novel polysaccharide controlled release system, Int. J. Pharm. 128(1996) 113–122.
[9] K. Chikhalikar, S. Moorkath, Carbopol Polymers: A versatile range of polymers forpharmaceutical applications, Pharmabiz, 2002.
[10] N.K. Ebube, A.B. Jones, Sustained release of acetaminophen from a heterogeneousmixture of two hydrophilic non-ionic cellulose ether polymers, Int. J. Pharm. 272(2004) 19–27.
[11] V.R.S. Uppoor, Regulatory perspectives on in vitro (dissolution)/in vivo (bioavail-ability) correlations, J. Control Release 72 (2001) 127–132.
[12] T. Higuchi, Mechanism of sustained-actionmedication. Theoretical analysis of rate ofrelease of solid drugs dispersed in solid matrices, J. Pharm. Sci. 52 (1963) 1145–1149.
[13] R.W. Korsmeyer, R. Gurny, E.M. Doelker, P. Buri, N.A. Peppas, Mechanism of soluterelease from porous hydrophilic polymers, Int. J. Pharm. 15 (1983) 25–35.
[14] N.A. Peppas, Analysis of Fickian and non-Fickian drug release from polymers,Pharm. Acta Helv. 60 (1985) 110–111.
[15] A.W. Hixson, J.H. Crowell, Dependence of reaction velocity upon surface andagitation, Ind. Eng. Chem. 23 (1931) 923–931.
[16] M. Kopcha, N. Lordi, K.J. Tojo, Evaluation of release from selected thermosofteningvehicles, J. Pharm. Pharmacol. 43 (1991) 382–387.
[17] V. Ratsimbazafy, E. Bourret, C. Brossard, Drug release from matrix tablets andminitablets containing glycerides, Pharm. Ind. 58 (1996) 442–446.
[18] US FDA CDER guidlines: modified release solid oral dosage forms guidance: scale-up and postapproval changes: chemistry, manufacturing and controls, In VitroDissolution Testing and In Vivo Bioequivalance Documentation, US Department ofHealth, Food and Drug Administration, Centre for Drug Elaluation and Research(CDER), Rocville, MD, 1997.
[19] N.K. Ebube, A.H. Hikal, C.M. Wyandt, D.C. Beer, L.G. Miller, A.B. Jones, Sustainedrelease of acetaminophen from heterogeneous matrix tablets: influence ofpolymer ratio, polymer loading, and co-active on drug release, Pharm. Dev.Technol. 2 (1997) 161–170.
[20] D.W.A. Bourne, Pharmacokinetics, in: G.S. Banker, C.T. Rhodes (Eds.), ModernPharmaceutics, (Drugs and the Pharmaceutical Sciences), 4th, vol. 121, MarcelDekker, Inc., New York, 2002, pp. 67–92.
[21] J.G. Wagner, Biopharmaceutics and Relevant Pharmacokinetics, 1st, Drug Intelli-gence Publishers, Illinois, 1971.
[22] Guidance for the Industry, U.S. Department of Health and Human Services, US FDA,CDER, Rockville, MD, 2000.
[23] US FDA CDER guidance for the industry: extended release solid oral dosage forms:development, Evaluation and Application of In Vitro/In Vivo Correlations, USDepartment of Health, Food and Drug Administration, Centre for Drug Elaluationand Research (CDER), Rocville, MD, 1997.
[24] F. Langenbucher, Handling of computational in vitro/in vivo correlation problemsbyMicrosoft Excel: III. Convolution and deconvolution, Eur. J. Pharm. Biopharm. 56(2003) 429–437.
[25] USP 29–NF 24, (United States Pharmacopoeia 29 National Formulary 24), U. S.Pharmacopeial Convention, Rockville, MD, 2006.
[26] J.C. Callahan, G.W. Cleary, M. Elefant, Equilibrium moisture content of pharma-ceutical excipients, Drug Dev. Ind. Pharm. 8 (3) (1982) 355–369.
[27] C.G. Velazquez de la, J. Torres, M. Martin-Polo, Temperature effects on the moisturesorption isotherms for methylcellulose and ethylcellulose films, J. Food. Eng. 48(2001) 91–94.
[28] P. Kulvanich, P. Leesawat, V. Patomchaiviwat, Release characteristics of thematrices prepared from co-spray-dried powders of theophylline and ethylcellu-lose, Drug Dev. Ind. Pharm. 28 (2002) 729–739.
[29] C.G. Cameron, J.W. McGinity, Controlled-release theophyl through line tabletscontaining acrylic resins III: influence of filler excipients, Drug Dev. Ind. Pharm.13 (2) (1987) 303–318.
[30] R.N. Chilamkurti, J.B. Schwartz, C.T. Rhodes, Effect of addition of soluble andinsoluble drugs on the disintegration of microcrystalline cellulose and dicalciumphosphate dihydrate, Pharm. Acta Helv. 58 (9–10) (1983) 251–255.
[31] V. Michailova, S. Titeva, R. Kotsilkova, E. Krusteva, E. Minkov, Influence of hydrogelstructure on the processes of water penetration and drug release from mixedhydroxypropylmethyl cellulose/thermally pregelatinized waxy maize starchhydrophilic matrices, Int. J. Pharm. 222 (2001) 7–17.
[32] V. Dhopeshwarkar, J.L. Zatz, Evaluation of xanthan gum in the preparation ofsustained release matrix tablets, Drug Dev. Ind. Pharm. 19 (1993) 999–1017.
[33] J.M. Cardot, E. Beyssac, In vitro/in vivo correlation: scientific implications andstandardization, Eur. J. Drug Metab. Pharmacokinet. 18 (1993) 113–120.
