p.

1 [100%]

YAKUGAKU ZASSHI 127(8) 1281―1290 (2007)  2007 The Pharmaceutical Society of Japan 1281

―Regular Articles―

Formulation and Optimization of Sustained Release Matrix Tablet of

Metformin HCl 500 mg Using Response Surface Methodology

Uttam MANDAL, Veeran GOWDA, Animesh GHOSH, Senthamil SELVAN,

Sam SOLOMON, and Tapan Kumar PAL
Bioequivalence Study Centre, Department of Pharmaceutical Technology,
Jadavpur University, Kolkata700 032, India

(Received March 21, 2007; Accepted May 18, 2007)

The aim of the current study was to design an oral sustained release matrix tablet of metformin HCl and to optimize
the drug release proˆle using response surface methodology. Tablets were prepared by non-aqueous wet granulation
method using HPMC K 15M as matrix forming polymer. A central composite design for 2 factors at 3 levels each was
employed to systematically optimize drug release proˆle. HPMC K 15M (X1 ) and PVP K 30 (X2 ) were taken as the in-
dependent variables. The dependent variables selected were ％ of drug released in 1 hr (rel1 hr), ％ of drug released in 8
hrs (rel8 hrs ) and time to 50％ drug release (t50％ ). Contour plots were drawn, and optimum formulations were selected
by feasibility and grid searches. The formulated tablets followed Higuchi drug release kinetics and diŠusion was the
dominant mechanism of drug release, resulting in regulated and complete release within 8 hrs. The polymer (HPMC K
15M) and binder (PVP K 30) had signiˆcant eŠect on the drug release from the tablets ( p＜0.05). Polynomial mathe-
matical models, generated for various response variables using multiple linear regression analysis, were found to be
statistically signiˆcant ( p＜0.05). Validation of optimization study, performed using 8 conˆrmatory runs, indicated
very high degree of prognostic ability of response surface methodology, with mean percentage error (±S.D.) 0.0437±
0.3285. Besides unraveling the eŠect of the 2 factors on the in vitro drug release, the study helped in ˆnding the optimum
formulation with sustained drug release.

Key words―response surface methodology; sustained release; matrix tablet; hydroxypropyl methyl cellulose (HPMC
K 15M); polyvinyl pyrrolidone (PVP K 30)

dosage form, an important issue is to design an op-

INTRODUCTION
timized formulation with an appropriate dissolution
Introduction of matrix tablet as sustained release rate in a short time period and minimum number of
(SR) has given a new breakthrough for novel drug trials. Many statistical experimental designs have
delivery system (NDDS) in the ˆeld of pharmaceuti- been recognized as useful techniques to optimize the
cal technology. It excludes complex production process variables. For this purpose, a computer based
procedures such as coating and pelletization during optimization technique with a response surface
manufacturing and drug release rate from the dosage methodology (RSM) utilizing a polynomial equation
form is controlled mainly by the type and proportion has been widely used.713) DiŠerent types of RSM de-
of polymer used in the preparations. Hydrophilic signs include 3-level factorial design, central compo-
polymer matrix is widely used for formulating an SR site design (CCD), Box-Behnken design and D-op-
dosage form.14) timal design. Response surface methodology (RSM)
Hydroxypropyl methyl cellulose (HPMC) is the is used when only a few signiˆcant factors are in-
widely used hydrophilic polymer to prolong drug volved in optimization. The technique requires mini-
release due to its rapid hydration, good compression mum experimentation and time, thus proving to be
and gelling characteristics along with its ease of use, far more eŠective and cost-eŠective than the conven-
availability and very low toxicity. It regulates the tional methods of formulating sustained release
release of drug by controlling the swelling and cross- dosage forms.
linking.5,6) Metformin HCl is an orally administered bigu-
In the development of a sustained release tablet anide, which is widely used in the management of
type-2 diabetes, a common disease that combines

e-mail: tkpal_12＠yahoo.com defects of both insulin secretion and insulin action.14)
p.2 [100%]

1282 Vol. 127 (2007)

It improves hepatic and peripheral tissue sensitivity to Hydroxy propyl methyl cellulose (HPMC K 15M)
insulin without the problem of serious lactic acidosis was a gift sample received from M/S Colorcon Asia
commonly found with its analogue, phenformin. It Pvt. Ltd., Mumbai, India. Microcrystalline cellulose
has three diŠerent actions: it slows the absorption of (MCC) and PVP K 30 (polyvinyl pyrrolidone K 30)
sugar in our small intestine; it also stops our liver were purchased from S. D. Fine Chemicals Ltd.,
from converting stored sugar into blood sugar; and it Mumbai, India. Magnesium stearate and talc were
helps our body use our natural insulin more e‹cient- procured from Mohanlal Dayaram and Company,
ly. It is a hydrophilic drug and is slowly and incom- Hyderabad. All other chemicals/reagents used were
pletely absorbed from the gastrointestinal tract and of analytical grade, except for those used in HPLC
the absolute bioavailability of a single 500 mg dose is analysis, which were of HPLC grade.
reported to be 5060％.15) An obstacle to more suc- Preparation of Sustained Release Matrix Tablets
cessful use of metformin therapy is the high incidence Table 1 enlists the composition of diŠerent trial for-
of concomitant gastrointestinal symptoms, such as mulations prepared using varying amounts of HPMC
abdominal discomfort, nausea, and diarrhea that es- K 15M as release controlling polymer and PVP K 30
pecially occur during the initial weeks of treatment. as binder along with ˆxed quantity of talcum and
Also the compound has relatively short plasma elimi- magnesium stearate as lubricant. MCC was used as
nation half-life of 1.5 to 4.5 hrs.16,17) Side eŠects and ˆller. HPMC K 15M polymer at diŠerent ratio was
the need for administration two or three times per day blended with metformin HCl, MCC and PVP K 30 in
when larger doses are required can decrease patient a planetary mixer for 5 mins after passing all the
compliance. Sustained release formulations that materials through a mesh (1150 mm). Thereafter the
would maintain plasma levels of drug for 8 to 12 hrs powders were granulated with isopropyl alcohol,
might be su‹cient for once daily dosing for metfor- sieved using a mesh (100 mm) and dried at 50° C for
min. SR products are needed for metformin to about 2 hrs with residual moisture content of 2 to 3％
prolong its duration of action and to improve patient w/w. The dried granules were sized by a mesh (250
compliance.15,18) mm) and mixed with magnesium stearate and talc for
Fiona et al.19) of Colorcon Ltd., UK has described 2 mins. All granules were weighed ˆnally to adjust the
the method for preparation of metformin HCl 500 ˆnal weight of individual tablet considering its loss
mg extended release tablet by direct compression during operational handling. Granules thus obtained
method. But in commercial scale it creates problem of were compressed into 1150 mg tablets to average
powder ‰ow ability from hoper to compression hardness of 6 to 8 kg/sq.cm on an eight station rotary
machine followed by weight variation, content tablet machine (CIP Machineries Pvt. Ltd., Ah-
uniformity, hardness and friability due to poor inher- medabad, India) with 19.5x8.9 mm caplet tooling at
ent compressibility of metformin HCl. a rotational speed of 72 rpm.
SR microcapsules of metformin by ethylcellulose Experimental Design A central composite de-
had been described by Balan et al.17) where metfor- sign (CCD) with a＝1 was employed as per the stand-
min gave in vitro release for up to 22 hrs. But prepa- ard protocol.8,11) The amounts of HPMC K 15M
ration of microcapsules in commercial scale and op-
timization of drug release rate is troublesome. Defang
et al.16) had described the bilayer matrix tablet and os- Table 1. Composition of 500 mg Metformin HCl Sustained
Release Matrix Tableta)
motic pump tablet consisting metformin and glipizide
both as SR form. The aim of this investigation was to Ingredient Amount (mg)
develop a sustained release matrix tablet of metfor- Metformin HCl 500 mg
min HCl using HPMC K 15M by non-aqueous wet HPMC K 15M 240 to 480 mg
granulation method and optimize the formulation us- PVP K 30 50 to 150 mg
ing RSM. Magnesium stearate 5 mg
Talcum powder 5 mg
MATERIALS AND METHODS MCC qs to 1150 mg
a) qs: quantity su‹cient, HPMC K 15M: Hydroxypropyl methyl cellu-
Materials Metformin HCl was received from lose of K 15M viscosity grade, PVP K 30: Polyvinyl pyrrolidone of K 30
Deys Medical, Kolkata, India as donate sample. viscosity grade, MCC: Microcrystalline cellulose.
p.3 [100%]

No. 8 1283

Table 2. Factor Combinations as per the Chosen Experimen- Acetonitrile＝50: 50 (v/v)

tal Design
Detector: UV detection with 232 nm
Coded factor levels Loop size: 20 ml
Trial No.
X1 X2
I －1 －1 Tablets were also evaluated for hardness (n＝10),
II －1 0 friability (n＝10), weight variation (n＝20), and
III －1 1 thickness (n＝10).
IV 0 －1
V 0 0 Drug Release Study Drug release from 6 tablets
VI 0 1 of each formulation, in triplicate, was determined us-
VII 1 －1
ing the USP I (basket) apparatus (Electrolab, TDT
VIII 1 0
06P, USP XXIII) where 900 ml of 0.1 N HCl and
IX 1 1
X 0 0
phosphate buŠer of pH 6.8 were used as dissolution
XI 0 0 media maintained at 37° C (±0.5° C) at 100 rpm. The
XII 0 0 release rates from the tablets were conducted in a dis-
XIII 0 0 solution medium of 0.1 N HCl for 2 hrs and there-
Translation of coded levels in actual units after in phosphate buŠer of pH 6.8 for 6 hrs. 5 ml of
Coded level －1 0 1 aliquot were withdrawn at 1, 2, 4 and 8 hrs with
X1: HPMC K 15M (mg) 240 360 480 replacement of fresh media. Solution samples were
X2: PVP K 30 (mg) 50 100 150
analyzed by high performance liquid chromatography
(HPLC) method mentioned in earlier section. Drug
release proˆles were drawn using MS-Excel software
(X1 ) and PVP K 30 (X2 ) were selected as the factors, and the values of t50％ were obtained by interpolation
studied at 3 levels each. The central point (0, 0) was from Excel graph.
studied in quintuplicate. All other formulation and Drug Release Kinetics In order to propose a
processing variables were kept invariant throughout possible release mechanism, drug release from
the study. Table 2 summarizes an account of the 13 HPMC matrix tablets was ˆtted to the following equ-
experimental runs studied, their factor combinations, ations:
and the translation of the coded levels to the ex- Higuchi's21) equation: Q＝KHt1/2 ( 1)
perimental units employed during the study. ％ of Where, Q is the amount of drug release at time t, and
drug released in 1 hr (rel1 hr ) (Y1), ％ of drug KH is the Higuchi rate constant.
released in 8 hrs (rel8 hrs ) (Y2 ) and time to 50％ drug Koresmeyer et al.'s20) equation: Mt/M∝＝ktn (2)
release (t50％ ) (Y3 ) were taken as the response varia- Where, Mt is the amount of drug released at time t,
bles. M∞ is the amount of drug released after inˆnite time,
Tablet Assay and Physical Evaluation 20 Mt/M∞ is the fractional drug release percentage at
tablets were taken and crushed to powder with mortar time t, k is a constant related to the properties of the
and pestle. Exact amount of powder (average weight) drug delivery system, and n is the release exponent in-
was taken and diluted with methanol up to 200 ml of dicative of the drug release mechanism.
volumetric ‰ask. After sonication for 15 mins, solu- Optimum Release Proˆle Optimum release
tion was ˆltered through 0.45 mm ˆlter paper. The proˆle for once-daily SR formulation was calculated
total amount of drugs within the tablets was analyzed by the following equation22) using available phar-
after appropriate dilution of test solution by using the macokinetic data:23)
HPLC method as described below against the refer- Dt＝Dose(1＋0.693×t/t1/2 ) (3 )
ence solution of metformin pure powder prepared in Where, Dt＝total dose of drug; Dose＝dose of the im-
the same procedure. mediate release part; t＝time (hr) during which the
sustained release is desired (8 hrs); t1/2＝half-life of
Column: Hypersil BDS C18 (250x4.6 mm, 5 mm par- the drug (3 hrs).
ticle size) The optimum formulation was selected based on
Mobile phase: 10 m.mol phosphate buŠer of pH 6.0: the above equation so that it could attain complete
p.4 [100%]

1284 Vol. 127 (2007)

and controlled drug release. Upon ``trading oŠ'' vari- sayed content of drug in various formulations varied
ous response variables, the following maximizing between 97.65％ and 99.53％ (mean 98.66％).
criteria were adopted: rel1 hr＝28 to 30％; rel8 hr＝95 to Tablets weights varied between 1140.5 and 1160.3 mg
100％ and t50％＝2.1 to 2.2 hrs. (mean 1152.57 mg), thickness between 7.45 and 7.56
Optimization Data Analysis and Validation of Op- mm (mean 7.52 mm), hardness between 5.8 and 7.3
timization Model Various RSM computations for kg.cm2 (mean 6.2 kg cm2 ), and friability ranged be-
the current optimization study were performed em- tween 0.15％ and 0.42％ (mean 0.31％). Thus, all the
ploying Design Expert software (Design Expert trial physical parameters of the matrices were practically
version 7.0.3 State-Ease Inc, Minneapolis, MN). within control.
Polynomial models including interaction and quad- In vitro Drug Release Studies Dissolution sam-
ratic terms were generated for all the response varia- ples were analyzed by HPLC method described in
bles using multiple linear regression analysis ``MATERIALS AND METHODS'' section. Metfor-
(MLRA) approach. The general form of the MLRA min was eluted at 2.920 mins from the analytical
model is represented as the following equation: column used for the analysis of dissolution sample.
y＝b0＋b1x1＋b2x2＋b3x1x2＋b4x21＋b5x22 Table 3 lists various dissolution parameters computed
＋b6x1x22＋b7x21x2 (4) for all the matrix formulations. To know the mecha-
Where, b0 is the intercept representing the arithmetic nism of drug release from the trial formulations, the
average of all quantitative outcomes of 13 runs; b1 to data were treated according to Higuchi's21) (cumula-
b7 are the coe‹cients computed from the observed ex- tive percentage of drug released versus square root of
perimental response values of Y; and X1 and X2 are time) and Koresmeyer et al.'s20) (log cumulative per-
the coded levels of the independent variable(s). The centage of drug released versus log time) equations.
terms X1X2 and X2i (i＝1 to 2) represent the interac- In our experiments the in vitro release proˆles of drug
tion and quadratic terms, respectively. Statistical from all the formulations could be best expressed by
validity of the polynomials was established on the ba- Higuchi's21) equation as the plots showed high lineari-
sis of ANOVA provision in the Design Expert ty (R2: 0.992 to 0.999, with KH 30.51 to 38.52) as
Software. Subsequently, the feasibility and grid shown in Table 3. In the current study, the values of
searches were performed to locate the composition of release rate exponent (n), calculated as per the equa-
optimum formulations.13,24)
Two-dimensional (2-D) contour plots were con-
structed based on the model polynomial functions us- Table 3. Drug Release Parameters of Various Trial Formula-
ing Design Expert Software. These plots are very use- tions Prepared as per the Experimental Designa)

ful to see interaction eŠects on the factors on the Factor amount

Trial (mg) rel1 hr rel8 hr t50％
responses. n KH R2
No. (％) (％) (hr.)
Eight optimum checkpoints were selected based on X1 X2
the criteria from optimum formulation described I 240 50 35.21 100.15 1.62 0.4993 35.43 0.994
earlier by intensive grid search, performed over the II 240 100 34.17 100.21 1.90 0.5117 35.78 0.998
entire experimental domain, to validate the chosen ex- III 240 150 33.21 99.12 1.95 0.5145 35.20 0.998
perimental design and polynomial equations. The for- IV 360 50 32.35 99.16 1.98 0.5263 35.72 0.999
V 360 100 30.47 99.19 2.11 0.5513 36.45 0.998
mulations corresponding to these checkpoints were
VI 360 150 27.65 85.32 2.25 0.5259 30.51 0.992
prepared and evaluated for various response proper-
VII 480 50 29.56 99.11 2.61 0.5824 37.92 0.992
ties. Subsequently, the resultant experimental data of
VIII 480 100 25.25 80.19 2.45 0.5387 29.18 0.988
response properties were quantitatively compared IX 480 150 23.15 73.11 3.92 0.5314 26.20 0.996
with that of their predicted values. Also, linear regres- X 360 100 28.18 99.14 2.13 0.5874 37.58 0.998
sion plots between observed and predicted values of XI 360 100 30.41 98.47 2.10 0.5528 36.39 0.999
the response properties were drawn using MS-Excel, XII 360 100 28.75 99.31 2.15 0.5813 37.42 0.998
forcing the line through origin XIII 360 100 27.98 100.15 2.17 0.5998 38.52 0.999
a) X1: HPMC K 15M, X 2: PVP K 30, rel1 hr: Release in 1 hr, rel8 hr:
RESULTS AND DISCUSSION Release in 8 hrs, t50％ : Time to 50％ drug release, n : Release exponent ob-
tained from Koresmeyer et al. equation (M t/M ∝ ＝kt n), KH: Higuchi rate
Drug Content and Physical Evaluation The as- constant (Q＝kH t 1/2), R2: Regression coe‹cient of Higuchi equation.
p.5 [100%]

No. 8 1285

tion proposed by Koresmeyer et al.,20) ranged be-

tween 0.4993 and 0.5874 (Table 3). For matrix
tablets, an n value of near 0.5 indicates diŠusion con-
trol, and an n value of near 1.0 indicates erosion or
relaxation control. Intermediate values suggest that
diŠusion and erosion contribute to the overall release
mechanism.25,26) In our experiments the results of n
clearly indicated that the diŠusion is the dominant
mechanism of drug release from these formulations.
DiŠusion is related to transport of drug from the
dosage matrix into the in vitro study ‰uid depending
on the concentration of the hydrophilic polymer. As
gradient varies, the drug is released, and the distance
for diŠusion increases. This could explain why the
Fig. 1. Cumulative Metformin Release (％) versus Time Pro-
drug diŠuses at a comparatively slower rate as the dis- ˆles for Metformin HCl Matrix Formulations Prepared as
tance for diŠusion increases. per the Experimental Design
Each value represents the mean ±S.D., n＝18
Total amount of metformin released from all the
formulations up to 8 hrs ranged between 73.11％ and
100.21％ indicating incomplete drug release at higher Table 4. Analysis of Variance (ANOVA) for All Three
concentration of HPMC K 15M as well as PVP K 30. Responsesa)
Rate of drug release (until 8 hrs) tended to decrease
rel1 hr (Y1) rel8 hr (Y2) t50％ (Y3)
with increase in the content of either HPMC or PVP
Source F p-value F p-value F p-value
K 30. This is in agreement with literature ˆndings27,28)
Model 16.90 0.003 9.96 0.011 26.90 0.001
that the viscosity of the gel layer around the tablet in- X1 34.00 0.002 15.20 0.012 7.86 0.038
creases with increase in the hydrogel concentration, X2 9.44 0.028 7.24 0.043 19.00 0.023
thus limiting the release of active ingredient. The gel X 1X 2 4.16 0.097 11.80 0.019 12.50 0.017
formed during the penetration of dissolution media X12 0.42 0.545 4.00 0.102 6.96 0.064
into the matrix structure, consists of closely packed X22 1.20 0.324 1.14 0.334 3.68 0.113
swollen particles. With further increase in polymer X12X2 0.07 0.802 0.01 0.961 5.24 0.071
X1X22 0.32 0.594 1.06 0.350 15.00 0.012
amount, thicker gel forms inhibiting dissolution me-
dia penetration more strongly, resulting in signiˆcant a) Signiˆcant eŠect ( p value＜0.5) of factors on individual responses
are shown in bold, rel1 hr: Release in 1 hr, rel8 hr: Release in 8 hrs, t50％ :
reduction in the values of rel8 hr indicating slower drug Time to 50％ drug release, X1: HPMC K 15M, X2: PVP K 30.
release.
The values of t50％ enhanced markedly from 1.62
hrs, observed at low levels of both the variables, to as tions showed quite regulated drug release from 4 hrs
high as 3.92 hrs, observed at high levels of both the onwards.
variables. This ˆnding indicated considerable release RSM Optimization Results
retarding potential of the polymer and binder. Mathematical Modeling Mathematical relation-
Figure 1 exhibits the mean (±S.D.) cumulative ships generated using MLRA for the studied response
metformin release (％) versus time proˆles obtained variables are expressed as Eqs. 5 to 7.
for various trial formulations, prepared as per CCD. y1＝29.2－4.46x1－2.35x2－1.10x1x2＋0.422x21
The formulations with lower levels of polymer and ＋0.712x22＋0.532x1x22＋0.248x21x2 (5)
binder exhibited initial burst in drug release. This y2＝97.9－10.00x1－6.92x2－6.24x1x2－4.38x21
result could be attributed to the dissolution of drug －2.34x22＋3.25x1x22＋0.163x21x2 (6)
present initially at the surface of the matrices and y3＝2.08＋0.275x1＋0.135x2＋0.245x1x2
rapid penetration of dissolution media to the matrix ＋0.220x21＋0.160x22＋0.465x1x22＋0.275x21x2
structure. However, the formulations showed little (7)
burst eŠect at higher polymer levels, ratifying better For estimation of signiˆcance of the model, the
substance of drug release. Overall, all the formula- analysis of variance (ANOVA) was determined as
p.6 [100%]

1286 Vol. 127 (2007)

per the provision of Design Expert Software (Table tude of the main eŠects signify the relative in‰uence
4). Using 5％ signiˆcance level, a model is considered of each factor on the response. The values obtained
signiˆcant if the p value (signiˆcance probability for main eŠects of each factor from Eqs. 5 to 7 reveal
value) is less than 0.05. From the p values presented that HPMC K 15M, individually, has rather more
in Table 4, it can be concluded that for all four pronounced eŠect on all response values. At a given
responses, the cross-product contribution (X1X2 ) set factor levels, however, these higher-order poly-
and quadratic contributions (X21, X22, X21X2 and X1 nomials yield results as the net eŠect of all the
X22 ) of the model was not signiˆcant. But the linear coe‹cient terms contained in the polynomial.
contribution (X1 and X2) for all three responses is Response Surface Analysis Figures 2, 3 and 4
signiˆcant (＜0.05). are the two-dimensional contour plots for the studied
The polynomial equations comprise the coe‹cients response properties viz rel1 hr, rel8 hr and t50％.
for intercept, ˆrst-order main eŠects, interaction Figure 2 exhibits that rel1 hr vary in a nonlinear
terms, and higher order eŠects. The sign and magni- fashion, but in a descending pattern with an increase

Fig. 2. Contour Plot Showing the Relationship between Various Levels of Polymer (HPMC K 15M) and Binder (PVP K 30) on
Drug Release in 1 hr

Fig. 3. Contour Plot Showing the Relationship between Various Levels of Polymer (HPMC K 15M) and Binder (PVP K 30) on
Drug Release in 8 hrs
p.7 [100%]

No. 8 1287

Fig. 4. Contour Plot Showing the Relationship between Various Levels of Polymer (HPMC K 15M) and Binder (PVP K 30) on
Time to Release 50％ Drug Release

Table 5. Composition of the Checkpoint Formulations, the Predicted and Experimental

Values of Response Variables, and Percentage Prediction Errora)

Composition: Response Experimented Predicted Percentage

HPMC K 15M：PVP K 30 variable value value error
rel1 hr 29.05 29.10 －0.172
336：130 rel8 hr 95.73 95.30 0.449
t50％ 2.13 2.12 0.469
rel1 hr 28.75 28.60 0.522
405.72：83.15 rel8 hr 96.27 96.50 －0.239
t50％ 2.17 2.17 0
rel1 hr 29.81 29.80 0.034
399：68.75 rel8 hr 99.11 99.30 －0.192
t50％ 2.15 2.16 －0.465
rel1 hr 28.83 28.90 －0.243
366：102.50 rel8 hr 97.19 97.10 0.093
t50％ 2.10 2.10 0
rel1 hr 28.85 28.80 0.173
348：123.10 rel8 hr 95.79 95.40 0.407
t50％ 2.12 2.13 －0.472
rel1 hr 29.27 29.20 0.239
392.16：80.35 rel8 hr 97.98 98.10 －0.123
t50％ 2.13 2.13 0
rel1 hr 28.79 28.70 0.313
354：117.50 rel8 hr 95.95 95.80 0.156
t50％ 2.13 2.13 0
rel1 hr 30.20 30.10 0.331
390：70 rel8 hr 99.01 99.70 －0.697
t50％ 2.15 2.14 0.465
a) rel1 hr: Release in 1 hr, rel8 hr: Release in 8 hrs, t50％ : Time to 50％ drug release, Percentage error (mean±
S.D.) 0.0437±0.3285.

in the amount of polymer and binder. It also shows In contrast to the results of drug release in 1 hr,
that HPMC K 15M has a comparatively greater in- contour plot for drug release in 8 hrs (Fig. 3) reveal
‰uence on the response variable than PVP K 30. that rel8 hr varies in somewhat linear fashion with in-
p.8 [100%]

1288 Vol. 127 (2007)

crease of polymer and binder contents. However, the Validation of RSM Results For all of the 8
eŠect of HPMC K 15M seems to be more pronounced checkpoint formulations, the results of the physical
as compared with that of PVP K 30. evaluation and tablet assay were found to be within
Figure 4 exhibits that time to 50％ drug release limits. Table 5 lists the compositions of the check-
(t50％ ) vary in a nonlinear manner, but in a ascending points, their predicted and experimental values of all
pattern with an increase in the amount of each varia- the response variables, and the percentage error in
bles. But at the higher amount of HPMC K 15M and prognosis.
PVP K 30 the contour lines turned to be linear. Figure 5 (A, C and E) shows linear correlation

Fig. 5. Linear Correlation Plots (A, C, E) between Observed and Predicted Values and the Corresponding Residual Plots (B, D, F)
for Various Variables
p.9 [100%]

No. 8 1289

plots between the observed and predicted response Pharm. Dev. Tech., 4, 313324 (1999).
variables, and the residual plots [Fig. 5 (B, D and 4) Lee B. J., Ryu S. G., Cui J. H., Drug Dev.
F)] showing the scatter of the residuals versus ob- Ind. Pharm., 25, 493501 (1999).
served values. 5) Chien Y. W., ``Novel Drug Delivery Sys-
Upon comparison of the observed responses with tems,'' 2nd ed., Marcel Dekker, New York,
that of the anticipated responses, the prediction error 1992.
varied between －0.697％ and 0.522％ (mean ±S.D. 6) Ravi Kumar M. N. V., Kumar N., Drug Dev.
as 0.0437 ±0.3285). The linear correlation plots Ind. Pharm., 27, 130 (2001).
drawn between the predicted and observed responses 7) Dave B. S., Amin A. F., Patel M. M., AAPS
demonstrated high values of r2 (ranging between Pharm. Sci. Tech., 5, 34 (2004).
0.9803 and 0.9900 excluding 0.8833 for t50％ ), indicat- 8) Singh B., Kumar R., Ahuja N., Crit. Rev.
ing excellent goodness of ˆt (p＜0.05). Relatively less Ther. Drug Carrier Syst., 22, 27105 (2005).
magnitudes of r2 observed with t50％ (0.8833) could be 9) Singh B., Ahuja N., Int. J. Pharm., 195, 247
attributed to the limitation of software (Design Ex- 248 (1999).
pert trial version 7.0.3 State-Ease Inc., Minneapolis, 10) Singh B., Dahiya M., Saharan V., Ahuja N.,
MN) to predict t50％ up to two decimal points only as Crit. Rev. Ther. Drug Carrier Syst., 22, 215
well as indirect estimation of observed t50％ values 293 (2005).
through interpolation techniques. 11) Singh B., Mehta G., Kumar R., Bhatia A., A-
huja N., Katare O. P., Curr. Drug Deliv., 2,
CONCLUSION
143153 (2005).
Controlled drug release following Higuchi kinetics 12) Aberturas M. R., Molpeceres J., Guzman M.,
attained in the current study indicates that the Garcia F., J. Microencapsul., 19, 6172
hydrophilic matrix tablet of metformin, prepared us- (2002).
ing HPMC K 15M and PVP K 30, can successfully be 13) Singh B., Ahuja N., ``Response Surface Op-
employed as once-a-day oral controlled release drug timization of Drug Delivery System, Progress
delivery system. Both the polymer and binder plays in Controlled and Novel Drug Delivery Sys-
major role for the sustained release of metformin. tems,'' eds. by Jain N. K., New Delhi, 2004.
However, appropriate balancing between various lev- 14) Stith B. J., Goalstone M. L., Espinoza R.,
els of the polymer and binder may contribute better Mossel C., Roberts D., Wiernsperger N. En-
results. High degree of prognosis obtained using RSM docrinology, 137, 29902999 (1996).
corroborates that a 2-factor CCD is quite e‹cient in 15) Dunn C. J., Peters D. H., Drugs, 49, 721749
optimizing drug delivery systems that exhibit non- (1995).
linearity in response(s). 16) Defang O., Shufang N., Wei L., Drug Dev.
Ind. Pharm., 31, 677685 (2005).
Acknowledgements The authors are thankful 17) Balan G., Timmins P., Greene D. S., Marathe
to Deys Medical, Kolkata, India for supplying the gift P. H., J. Pharm. Sci., 8, 11761185 (2001).
samples of metformin HCl .The authors also ac- 18) Montvale N. J., ``Physicians' Desk Refer-
knowledge All India Council for Technical Education ence,'' 53rd ed., Medical Economics Co.,
(AICTE), New Delhi, India to carry out this project 1999.
through their Grant No. 110/NDF (PG)/JU (02)/ 19) Fiona P., Marina L., Ali R. S., ``Investigation
200405. of a Directly Compressible Metformin HCl
500 mg Extended Release Formulation based
REFERENCES
on Hypromellose,'' Poster Reprint, Con-
1) Vidyadhara S., Rao P. R., Prasad J. A., Indi- trolled Release Society Annual Meeting, 2005.
an J. Pharm. Sci., 66, 188192 (2004). 20) Koresmeyer R. W., Gurny R., Doelker E. M.,
2) Reddy K. R., Mutalik S., Reddy S., AAPS Buri P., Peppas N. A., Int. J. Pharm., 15, 25
Pharm. Sci. Tech., 4, 19 (2003). (1983).
3) Mohammed A. D., James L. F., Michael H. 21) Higuchi W. I., J. Pharm. Sci., 51, 802804
R., John E. H., RajabiSiahboomi A. R., (1962).
p.10 [100%]

1290 Vol. 127 (2007)

22) Rawlins E. A. ``Bentley's Text Book of Phar- 26) Peppas N. A., Pharm. Acta Helv., 60, 110
maceutics,'' Cassell and Collier Macmillan, (1985).
London, 1977. 27) Ford J. L., Rubinstein M. H., Hogan J. E.,
23) Defang O., Shufang N., Wei L., Drug Dev. Int. J. Pharm., 24, 327338 (1985).
Ind. Pharm., 31, 677685 (2005). 28) Vazques M. J., Perez-Marcos B., Gomez-
24) Singh B., Ahuja N., Drug Dev. Ind. Pharm., Amoza J. L., Martinez-Pacheco R., Souto C.,
28, 431442 (2002). Concheiro A., Drug Dev. Ind. Pharm., 18,
25) Fassihi R. A., Ritschel W. A., J. Pharm. Sci., 13551375 (1992).
82, 750754 (1993).