European Journal of Pharmaceutical Sciences 37 (2009) 434441

Contents lists available at ScienceDirect

European Journal of Pharmaceutical Sciences
j our nal homepage: www. el sevi er . com/ l ocat e/ ej ps
Application of biorelevant dissolution tests to the prediction of in vivo
performance of diclofenac sodium from an oral modied-release
pellet dosage form
Ekarat Jantratid
a,
, Vincenzo De Maio
b
, Emanuela Ronda
b
, Valentina Mattavelli
b
,
Maria Vertzoni
c
, Jennifer B. Dressman
a
a
Institute of Pharmaceutical Technology, Johann Wolfgang Goethe University, Max-von-Laue-Str. 9, 60438 Frankfurt am Main, Germany
b
Eurand SpA, Milan, Italy
c
Faculty of Pharmacy, National and Kapodistrian University of Athens, Athens, Greece
a r t i c l e i n f o
Article history:
Received 28 January 2009
Received in revised form 23 March 2009
Accepted 27 March 2009
Available online 5 April 2009
Keywords:
Biorelevant dissolution tests
Bio-Dis
Flow-through cell
In vitroin vivo correlations
Food effects
Diclofenac sodium
a b s t r a c t
In vitro biorelevant dissolution tests enabling the prediction of in vivo performance of an oral modied-
release (MR) dosage formwere developedinthis study. Invitro dissolutionof MRdiclofenac sodiumpellets
containing 100mg active ingredient was evaluated under simulated pre- and postprandial conditions
using USP Apparatus 3 (reciprocating cylinder, Bio-Dis) and 4 (ow-through cell) and results compared
with compendial methods using USP Apparatus 1 (basket) and 2 (paddle). In vivo, the effects of food
on the absorption of diclofenac sodium from the pellet dosage form were investigated by administering
the product to 16 healthy volunteers pre- and postprandially in a crossover-design study. The in vitro
results were compared with the in vivo data by means of Level A in vitroin vivo correlation (IVIVC) and
Weibull distribution analysis. The compendial dissolution tests were not able to predict food effects. The
biorelevant dissolution tests predicted correctly that the release (and hence absorption) of diclofenac
sodium would be slower in the fed state than in the fasted state. No signicant differences in extent of
absorption due to changes in extent of release were predicted or observed. The results demonstrate good
correlations between in vitro drug release and in vivo drug absorption in both pre- and postprandial states
using the biorelevant dissolution test methods.
2009 Elsevier B.V. All rights reserved.
1. Introduction
For some years attempts have been made to minimize the num-
ber of in vivo studies necessary to approve newdrug products. One
of the approaches currently used is the in vitroin vivo correla-
tion (IVIVC) (Uppoor, 2001; Emami, 2006; US FDA, 1997b). With
respect to the IVIVC concept, in vitro (mainly dissolution) tests are
applied as a tool to predict drug product performance in vivo (US
FDA, 1997b; USP 30, 2008). The best candidates for IVIVC analysis
are products for which dissolution is the overall rate-determining
step to drug absorption. There are many studies demonstrating
different aspects of IVIVC for products containing BCS Class II com-
pounds housed in immediate-release (IR) dosage forms (Amidon et
al., 1995; Dressman and Reppas, 2000; Wei and Lbenberg, 2006;
Sunesen et al., 2005) and for modied-release (MR) dosage forms
in general (Kortejrvi et al., 2002; Balan et al., 2001; Frick et al.,

Corresponding author. Tel.: +49 69 798 29685; fax: +49 69 798 29694.
E-mail address: jantratid@em.uni-frankfurt.de (E. Jantratid).
1998; Takka et al., 2003; Wingstrand et al., 1990; Abrahamsson
et al., 1994). Some of these studies employed dissolution media
describedinthe pharmacopeias (so-calledcompendial approach)
(Kortejrvi et al., 2002; Balan et al., 2001; Frick et al., 1998; Takka et
al., 2003), while others added synthetic surfactants to compendial
media (Wingstrand et al., 1990; Abrahamsson et al., 1994; Rossi et
al., 2007). As these conditions do not comprehensively represent
the gastrointestinal (GI) tract environment, it can be inferred that
the results can only be interpreted on an empirical basis.
As a part of a general drive to develop predictive in vitro models,
biorelevant media were proposed and have evolved over the last
decade as a tool for in vitro biorelevant dissolution tests (Dressman
et al., 1998; Galia et al., 1998; Nicolaides et al., 1999; Vertzoni et
al., 2004). Recently, the media have been updated to more nearly
represent both the pre- and postprandial states in the proximal gut
(Dressman et al., 2007; Lue et al., 2008; Jantratid et al., 2008b). The
media compositions have beenne-tunedaccordingtothe phase of
digestion both in the stomach and the upper small intestine, since
these factors can dramatically affect drug solubility and dissolution
rate. By employing these biorelevant media it should be possible to
0928-0987/$ see front matter 2009 Elsevier B.V. All rights reserved.
doi:10.1016/j.ejps.2009.03.015
E. Jantratid et al. / European Journal of Pharmaceutical Sciences 37 (2009) 434441 435
predict the effects of foodondrug exposure in vivo, whichcandiffer
by as much as one or two orders of magnitude.
One limitation of the compendial approaches to simulating the
in vivo release proles of dosage forms with MR properties is that
results are typically run in one medium at a time. Only in a lim-
ited number of cases (most notably enteric coated products) does
the compendial method call for a change in the composition of the
mediumduring the test (USP 30, 2008). This is obviously incontrast
to the changing environment to which the dosage forms are gener-
ally exposed as they pass through the GI tract. Additionally, for MR
drug products, the release of drug may be inuenced by the his-
tory of the dosage form after ingestion. For example, swelling of
polymer components under gastric conditions may inuence the
subsequent release under intestinal conditions or lead to such a
reduction in mechanical resistance that the dosage form breaks up
upon passage through the ileocecal valve. Thus, it is desirable to
set up the in vitro release test conditions for MR dosage forms in
sucha way (i.e. witha series of media inone experiment) that these
effects can be observed and predicted.
The USP Apparatus 3 (reciprocating cylinder, Bio-Dis) and 4
(ow-through cell) offer the advantages of determining release
from the dosage form under various, consecutive conditions simu-
lating the GI physiology. The release experiments performed with
Bio-Dis and ow-through cell can be set up with a series of dissolu-
tion media in one single run, thus making it possible to mimic the
history of the dosage formas it passes through the GI tract and to
generate an IVIVC on an a priori basis.
Diclofenac sodium, a non-steroidal anti-inammatory drug,
was used as a model compound in this study. It was formu-
lated into capsules containing MR pellets consisting of 100mg
active pharmaceutical ingredient (API) per dosage unit using
ammonio methacrylate copolymer type A (NF) (USP 30, 2008)
as a release-modifying agent. The release of diclofenac sodium
from MR pellet dosage forms was compared using Bio-Dis and
ow-through cell with USP Apparatus 1 (basket) and 2 (paddle
assembly). Food effects on the release of diclofenac sodium, and
hence its absorption, were predicted using biorelevant media sim-
ulating the fasted and fed states in the human GI tract (Vertzoni
et al., 2005; Fotaki et al., 2005; Dressman et al., 2007; Jantratid
et al., 2008b). In vitro results were compared with pharmacoki-
netic data obtained in the fed and fasted human subjects. Level
A IVIVC (USP 30, 2008; Levy and Hollister, 1964; Levy et al.,
1965) and RRSBW (Weibull) distribution analysis (Langenbucher,
2005) were applied to assess predictability of the in vitro
model.
2. Materials and methods
2.1. Materials
Diclofenac sodium capsules (lot P200750226) were manufac-
tured by Eurand SpA. Acetonitrile and methanol were of gradient
grade and obtained from Merck KGaA, Darmstadt, Germany. Egg
phosphatidylcholine (Lipoid E PC

, 97.9% pure, lot 108027-1/045)

was purchased from Lipoid GmbH, Ludwigshafen, Germany. Glyc-
eryl monooleate (GMO, Rylo MG19 Pharma

, 99.5%monoglyceride,
lot 4010380689) was a gift from Danisco Specialities, Brabrand,
Denmark. Hydrochloric acid (37%, fuming) was purchased from
Mallinckrodt Baker, Milan, Italy. Calcium chloride (97% pure, anhy-
drous, lot 1309867), ortho-phosphoric acid (85%) and pepsin (Ph.
Eur., 0.51U/mg, lot 1296935) were obtained from Fluka Chemie
AG, Buchs, Switzerland. Maleic acid (99% pure, lot 056K5473) was
purchased from SigmaAldrich Chemie GmbH, Steinheim, Ger-
many. Sodium oleate (82.7% pure, lot 61720) was obtained from
Riedel-deHan, Seelze, Germany. Sodiumtaurocholate(NaTC, >99%
pure, lot 2006040099) was purchased from Prodotti Chimici e Ali-
mentari SpA, Basaluzzo, Italy. Ammonium dihydrogen phosphate,
dichloromethane, sodiumdihydrogenphosphatemonohydrateand
tetrahydrofuran(THF) were obtainedfromMerckKGaA, Darmstadt,
Germany. Isopropyl alcohol and n-hexane were purchased from
BDH, Milan, Italy. Glacial acetic acid, sodium acetate trihydrate,
sodium chloride, sodium dihydrogen phosphate monohydrate and
sodiumhydroxidepellets wereall of analytical gradeandpurchased
from Carlo Erba Reagenti SpA, Milan, Italy. Pancreatin powder (72
USP lipase Units/mg, lot 1279-0057) was obtained from Scientic
Protein Laboratories LLC, WI, USA.
2.2. Quantitative analysis of diclofenac sodium
The amounts of diclofenac dissolved in the samples were
analyzed using a validated isocratic reversed phase HPLC sys-
tem. The system consisted of a Waters Alliance 2695 quaternary
pump, a Waters Alliance 2695 autosampler, a Waters 996 Pho-
todiode Array Detector (Milford, MA, USA), an integrator (CR5A
Shimadzu, Shimadzu, Kyoto, Japan) and an RP-18-e column, 5m,
125mm4mm LiChrospher 100 connected with a RP-18-e guard-
column, 5m, 4mm4mm LiChrospher 100. The mobile phase
consisted of 63% of 0.05M ammonium phosphate bufferpH 5.0,
29.5% acetonitrile and 7.5% tetrahydrofuran (by volume). The ow
rate was set at 1.2mL/minresulting ina runtime of 15minper sam-
ple. The injection volume was 20L. The detection wavelength was
set at 275nm.
The amounts of diclofenac in the plasma samples were quan-
tied by extracting the acidied (with 0.5N hydrochloric acid
solution) plasma with an n-hexane/isopropyl alcohol mixture
(95:5). The organic layer was evaporated under nitrogen stream
at 50

C. The dried residues were reconstituted with acetoni-

trile/water mixture (1:1) and theninjected into the HPLC. The HPLC
system consisted of a Spectra Physics SP 8800 ternary pump, a
Shimatzu SIL-9A autosampler, a Spetra Physics SP 8480 XR scan-
ning spectrophotometer, a PCAX2 (Epson) withOmega 2 Analytical
Workstation software integrator (Perkin Elmer) and an HS 5C8 col-
umn, 5m, 15mm4.6mmI.D. (Perkin Elmer) connected with an
Upchurch Perisorb RP-18 guardcolumn, 3040m, 3cm2.1mm
I.D. (Merck). The mobile phase consisted of 47% acetonitrile and
53% of 0.4% ortho-phosphoric acid (by volume). The ow rate was
set at 1.0mL/min resulting in a run time of 10min per sample. The
injection volume was 50L. The detection wavelength was set at
275nm.
2.3. Modied-release formulations
Diclofenac sodium pellets (with a pellet size of approximately
2mm and the average API content of 100.7%) were manufactured
and lled in hard gelatin capsules at a label strength of 100mg of
diclofenac sodiumper dosage unit. Ammonio methacrylate copoly-
mer type A (NF) (USP 30, 2008), a pH-independent, permeable
polymer, was used as a release-modifying agent.
2.4. Dissolution testing
Quality control (QC) compendial methods using the basket and
paddle and biorelevant methods using the reciprocating cylinder
(Bio-Dis III Extended Release Tester, Varian Inc., CA, USA) and the
ow-throughcell (Sotax S.r.l., Bergamo, Italy) were usedto evaluate
the dissolution behavior of diclofenac sodium from the MR pellet
dosage form.
436 E. Jantratid et al. / European Journal of Pharmaceutical Sciences 37 (2009) 434441
Table 1
Biorelevant dissolution media simulating the preprandial conditions in the gastrointestinal tract.
Medium FaSSGF
a
New-FaSSIF
b
Half-FaSSIF FaSSIF-sans SCoF
c
Bile secretions
- NaTC 80M 3.0mM 1.5mM - -
- Lecithin 20M 0.2mM 0.2mM - -
Enzyme
- Pepsin 0.1mg/mL - - - -
- Pancreatin (lipase Unit/mL) - - - - -
pH 1.6 6.5 7.0 7.5 5.8
Osmolality (mOsm/kg) 120.7 180 270 270 295
Buffer capacity (mEqL
1
pH
1
) - 10 10 10 29
Abbreviations: FaSSGFFasted State Simulated Gastric Fluid; FaSSIFFasted State Simulated Intestinal Fluid; SCoFSimulated Colonic Fluid; NaTCsodium taurocholate.
a
Vertzoni et al. (2005).
b
Dressman et al. (2007) and Jantratid et al. (2008a,b).
c
Fotaki et al. (2005).
Table 2
Biorelevant dissolution media simulating the postprandial conditions in the gastrointestinal tract.
Medium FeSSGF
a
New-FeSSIF
b
Half-FeSSIF FaSSIF-sans
c
SCoF
d
Bile secretions
- NaTC - 7.5mM 3.0mM - -
- Lecithin - 2.0mM 1.0mM - -
Lipolytic products
- GMO - 5.0 - - -
- Sodium oleate - 0.8 - - -
Enzyme
- Pepsin - - - - -
- Pancreatin (lipase Unit/mL) - 100 - - -
pH 5.0 5.8 6.5 7.5 5.8
Osmolality (mOsm/kg) 400 390 270 270 295
Buffer capacity (mEqL
1
pH
1
) 25 25 10 10 29
Abbreviations: FeSSGFFed State Simulated Gastric Fluid; FeSSIFFed State Simulated Intestinal Fluid; GMOglyceryl monooleate
a
Jantratid et al. (2008a,b). This medium contains 50% full fat (3.5%) UHT-milk.
b
Dressman et al. (2007).
c
The conditions during the late postprandial state in the upper small intestine are similar to the preprandial state. Therefore, FaSSIF-sans was used for simulating both
conditions.
d
Fotaki et al. (2005).
2.4.1. Quality control method
The QC test conditions for the basket method consisted of
900mL phosphate buffer, pH 6.8 as a dissolution medium, the bas-
ket rotation speed of 50rpm, and a temperature of 37

C0.5

C.
The paddle method employed the same conditions as for the bas-
ket method, except for the paddle rotational speed, which was set
at 75 or 125rpm. Experiments were conducted in six replicates.
The sampling times were 30, 60, 120, 180 and 240min. Sampling
was performedautomaticallythroughthesamplingdevice. Thevol-
ume withdrawn was approximately 5mL for each sampling time
point. The samples were ltered through a 0.45m polytetrauo-
roethylene (PTFE) lter and then analyzed spectrophotometrically
at 275nm.
2.4.2. Biorelevant methods
2.4.2.1. Bio-Dis method. A biorelevant, pH-gradient method using
the Bio-Dis tester was applied to simulate release of diclofenac
sodium from the MR pellet dosage form in the GI tract pre- and
postprandially. The biorelevant media used in this study and the
dissolution test set-up are shown in Tables 13. The dissolution
experimental design was modied from that proposed previously
(Klein et al., 2005, 2008). The detailed compositions and media
preparation have mostly been described elsewhere (Vertzoni et al.,
2005; Fotaki et al., 2005; Dressman et al., 2007; Jantratid et al.,
2008b). The dissolution conditions consisted of a media volume
of 220mL per vessel with a dip rate of 10dpm. The top and bot-
tom mesh size was 405m (40 mesh). The sampling times for the
preprandial simulation were 30, 60, 105, 150, 210, 270, 360, 450,
540 and 750min and for the postprandial simulation 120, 165, 210,
270, 330, 450, 630 and 810min. The temperature in the vessels
was maintained at 370.5

C. Experiments were conducted in six

replicates. Samplingwas performedautomaticallyusingasampling
device (VanKel VK 8000 Dissolution Sampling Station, Varian Inc.,
Table 3
Preprandial and postprandial pH-gradient schedule for the biorelevant dissolution
tests of MR diclofenac sodium pellets using Bio-Dis method.
Segment of the GI tract pH-gradient preprandial Residence time (min)
Biorelevant medium pH
Stomach FaSSGF 1.6 60
Duodenum/Jejunum New-FaSSIF 6.5 45
Jejunum/Ileum Half-FaSSIF 7.0 45
Distal ileum FaSSIF-sans 7.5 120
Colon SCoF 5.8 480
Segment of the GI tract pH-gradient postprandial Residence time (min)
Biorelevant medium pH
Stomach FeSSGF 5.0 120
Duodenum/Jejunum New-FeSSIF 5.8 45
Jejunum/Ileum Half-FeSSIF 6.5 45
Distal ileum FaSSIF-sans 7.5 120
Colon SCoF 5.8 480
E. Jantratid et al. / European Journal of Pharmaceutical Sciences 37 (2009) 434441 437
CA, USA). The sample volume withdrawn was approximately 5mL.
With the exception of FeSSGF, the samples were ltered through
a 0.45m PTFE lter and then analyzed by HPLC. The amount of
drug released in the FeSSGF was indirectly determined by using
the innity point approach (Klein, 2005) by adding one more row
of vessels containing phosphate buffer, pH6.8to the series of media
set-up in the postprandial state. After the release was tested in the
rst verows, pellets wereexposedtothis mediumuntil therelease
was exhausted (20dpm, 6h). By subtracting the cumulative release
in all vessels analyzed from the label strength of the dosage form
(100mg), the drug release in the rst row can be estimated.
2.4.2.2. Flow-through cell method. The ow-through dissolution
tester equippedwith22.6mmdiameter test cells andapistonpump
was used to evaluate the release of diclofenac sodium from the
pellets during exposure to the biorelevant, pH-gradient methods.
A 5mm-sized glass bead was placed in the tip of the cell. A total
of 1.7g of 1mm-sized glass beads were added above the 5mm
glass bead, while a glass ber (MNGF1, 0.7m pore size, 25mm
diameter, Machery-Nagel, Germany) was placed on the top of the
cell. During the experiment, the capsule was mounted on a holder.
Experiments were performed in triplicate at 370.5

C in dissolu-
tion media simulating the compositions of gastric, small intestinal
and ascending colonic contents in the fasted and fed states (as
for the Bio-Dis experiments). Duration of exposure to the vari-
ous fasted and fed state simulating media and the corresponding
ow rates were adapted from Fotaki et al. (2005). Fluid exiting the
ow-cell was collected in a volumetric cylinder. The cylinders were
exchanged every 20min up to 420min.
The content of diclofenac sodium in each cylinder was ana-
lyzed by HPLC. Except for FeSSGF, the samples collected from the
experiments were immediately injected into the HPLC systemafter
ltration. Incase of FeSSGF, 1mL of the collectedsamples was trans-
ferred to a tube and 2mL of acetonitrile was added. After vortexing
for 1min, the sample was centrifuged for 10min at 4000rpm. The
clear supernatant was then injected into the HPLC system.
2.5. Comparative bioavailability studies
The MR diclofenac sodium pellets were given to 16 healthy
volunteers [nine males and seven females, mean age24.9 years
old (range 2033 years old); mean body weight65kg (range
5283kg); mean height173cm(range 164185cm)] in the fasted
and fed states on a randomized crossover basis with a wash-out
period of one week. The study was conducted following the stan-
dard procedure of the US FDA for the assessment of food effects on
drug absorption (US FDA, 2002) and the recommendations of the
Declaration of Helsinki. The study protocol was approved by the
local ethics committee prior to the beginning of the study. Blood
samples were collected every hour up to 12h and the last sam-
ples were collectedat 24h. The diclofenac concentrations inplasma
were determined using the aforementioned HPLC method.
2.6. Analysis of in vitro dissolution data
Differences in the in vitro dissolution proles were assessed
using the model-independent approach based on the similarity
factor (f
2
) as follows (US FDA, 1997a):
f
2
= 50 log

1 +
1
n

n
t=1
(R
t
T
t
)
2

0.5
100

(1)
where n is the number of time points, R
t
is the dissolution value
of the reference at time t, and T
t
is the dissolution value of the
test at time t. The f
2
is basically a measurement of the similarity
in the percent (%) drug dissolution between the two curves. Values
of 50 or above (50100) ensure similarity (difference 10%) of the
curves.
2.7. Analysis of in vivo pharmacokinetic data
Non-compartmental analysis was applied to the evaluation of
pharmacokinetic parameters using WinNonlin

Professional Edi-
tion version 4.1 software (Pharsight Corporation, Mountain View,
CA, USA) and Microsoft Ofce Excel

2003 (Microsoft Corporation,

Redmond, WA, USA). Students pairedt-test was usedtostatistically
interpret the differences between the area under the plasma drug
concentrationtime curves (AUC
0t
) and the maximum plasma
drug concentration(C
max
) after administrationinthe fedandfasted
states, while Wilcoxons signed-rank test was used for the T
max
comparison. A probability level of 0.01 was applied to all statistical
analyses.
2.8. In vitroin vivo correlations
The plasma drug concentrationtime proles obtained in the
pre- and postprandial states were deconvoluted to the fraction
drug absorbed (F
a
) proles using WinNonlin

software. The unit

impulse response was determined from the literature plasma
drug concentrationtime data obtained following the intravenous
administration of 50mg diclofenac sodiumto seven healthy female
volunteers (Willis et al., 1979). The fraction drug dissolved (F
d
) pro-
les were taken directly from the dissolution data.
The F
a
vs F
d
curve comparisons were determined using the
Weibull distribution as described by the following equation:
W
t
= W
max
[1 e
[(t)/
d
]

] (2)
where W
t
is the fraction of drug dissolved/absorbed at time t, W
max
is the maximum cumulative fraction dissolved/absorbed, is the
location parameter (the lag time before the onset of dissolution),
d
is the time parameter (provides information about the overall rate
of the process), and is the shape parameter. The proles were t
with the SigmoidalWeibull distribution with a slight modica-
tion using SigmaPlot

software, version 10.0 (Erkrath, Germany).

Togenerate Level AIVIVCthe dissolutionresults fromthe QCand
biorelevant methods were plotted against the in vivo drug absorp-
tion data, i.e. the relationship between F
a
and F
d
values, in which
the time pairs in vitro (drug dissolution) and in vivo (drug absorp-
tion) are correlated (USP 30, 2008; Levy and Hollister, 1964; Levy
et al., 1965). No time scale corrections were made for the prole
comparison.
3. Results and discussion
3.1. Biorelevant dissolution media
Biorelevant dissolution media have been updated recently
(Dressman et al., 2007; Lue et al., 2008; Jantratid et al., 2008b) to
better simulatetheproximal humanGI conditions thantheir prede-
cessors, i.e. FaSSIF and FeSSIF (Galia et al., 1998). In fact, the present
study was performed during the development of the updated
biorelevant dissolution media (Jantratid et al., 2008b) and the
design of media compositions described herein was based mostly
on the recent in vivo data collected from human aspirates in the
pre- and postprandial states (Kalantzi et al., 2006). The media com-
positions, as tabulated in Tables 1 and 2, reect the physiological
changes and the on-going digestive processes along the human GI
tract in the fasted and fed states. Crucial parameters affecting drug
solubility and dissolution including (i) the levels of bile secretions,
438 E. Jantratid et al. / European Journal of Pharmaceutical Sciences 37 (2009) 434441
(ii) the presence of lipolytic products and enzymes, and (iii) the pH,
buffer capacity and osmolality, were all taken into consideration. In
this study, media representing the fasted and fed stomach (FaSSGF
and FeSSGF), the small intestine in the fasted (New-FaSSIF, Half-
FaSSIF, and FaSSIF-sans) and fed state (New-FeSSIF, Half-FeSSIF, and
FaSSIF-sans), and the ascending colon (SCoF) were developed or
adoptedfromtheliterature(Vertzoni et al., 2005; Fotaki et al., 2005;
Dressman et al., 2007; Jantratid et al., 2008b). FaSSIF-sans (pH 7.5)
was used for both the pre- and postprandial states to simulate the
lower part of the small intestine, where reuptake of the bile com-
ponents is virtually complete and pH is higher. Likewise, SCoF was
used as a dissolutionmediumto represent the conditions inhuman
colon in both the pre- and postprandial states (Fotaki et al., 2005).
3.2. Quantitative analysis of diclofenac sodium
Validation of the chromatographic method for diclofenac
sodiumassay fromthe in vitro tests showed that the limit of detec-
tion(LOD) was 0.09g/mL andthelimit of quantication(LOQ) was
0.3g/mL. The mean recovery of 99.0% was observed within the
concentration range of 15150g/mL. The accuracy and presicion
of the assay were less than 5%.
For the analysis of diclofenac sodium concentration in plasma,
the LOD was 2ng/mL and the LOQ was 10ng/mL. The mean recov-
ery of 99.7% was observed within the concentration range of
102000ng/mL. The accuracy and presicion of the assay were less
than 10%.
3.3. Dissolution testing
3.3.1. Quality control methods
Fig. 1A shows the dissolution proles of diclofenac sodiumfrom
the MR pellets using the QC methods. Basket and paddle methods
gave comparable dissolution results. Additionally, using the paddle
method, rotational speeds of 75 and 125rpmshowed no signicant
differences in the release rate of diclofenac sodiumformthe pellets
(f
2
value =67.3). Approximately 85% drug release was observed in
all test conditions within 120min.
3.3.2. Biorelevant methods
The dissolution proles of diclofenac sodium from the MR pel-
lets using the Bio-Dis and ow-through methods are shown in
Fig. 1B. In this study, the biorelevant dissolution tests were used
to simulate not only differences in GI uid compositions between
the pre- and postprandial states, as reected by the dissolution
media compositions, but also differences in the residence times
in the different parts of the GI tract, as reected by the passage of
the dosage formin the biorelevant dissolution test set-up (Table 3).
Depending on the dimension of the dosage units, e.g. monolithic vs
multiparticulate dosage forms, the residence times can be drasti-
cally different especially in the stomach (Hardy et al., 1993; Davis
et al., 1986; Coupe et al., 1991). For example, gastric emptying and
thus arrival at regions in the small intestine with higher pH values
is generally quicker in the fasted state than in the fed state. Further,
the gastric residence time is usually shorter for multiparticulates
than monoliths (Hardy et al., 1993; Davis et al., 1986; Coupe et al.,
1991). Based on the literature data, the residence times of the pel-
lets in different regions of the human GI tract in the fasted and fed
conditions (Hardy et al., 1993; Davis et al., 1986; Coupe et al., 1991)
were applied to the design of the biorelevant dissolution tests, as
described in Table 3.
The results in Fig. 1B show that under postprandial conditions
the release of diclofenac sodium from the pellets would be slower
than in an empty stomach. Weibull analysis (
d
and values;
Table 4) indicates an apparent difference between the dissolution
Fig. 1. Dissolution prole comparison of cumulative diclofenac sodium dissolu-
tion vs time proles using (A) the QC methods (basket and paddle); and (B) the
biorelevant methods in the fasted and fed state using the Bio-Dis and ow-through
apparatus.
characteristics in the fasted and fed states using either Bio-Dis
or ow-through cell. Owing to the weakly acidic properties of
diclofenac (pK
a
3.8) (OConnor and Corrigan, 2001), the release in
the fasted stomach in normal acid secretors is expected to be poor.
However, the gastric residence time is short and it can be expected
that the release would go up immediately after the dosage form
has been emptied into the duodenum. By contrast, after ingestion
of the meal, the gastric pH initially increases (Russell et al., 1993;
Table 4
Weibull parameters derived from the fraction drug absorbed and the fraction drug
dissolved proles.
Weibull
parameters
Fain vivo
a
F
d
Bio-Dis
b
F
d
Flow-through
c
Fasted
d
(min) 243.7 167.3 193.2
1.7 2.8 2.9
Fed
d
(min) 420.5 249 289
2.6 4.3 2.7
a
Fain vivofraction of drug absorbed deconvoluted from the in vivo plasma
proles.
b
F
d
Bio-Disfraction of drug dissolved obtained from Bio-Dis method.
c
F
d
Flow-throughfraction of drug dissolved obtained from ow-through cell
method.
E. Jantratid et al. / European Journal of Pharmaceutical Sciences 37 (2009) 434441 439
Fig. 2. (A) Plasma diclofenac sodium concentrationtime proles obtained follow-
ing a single oral administration of modied-release diclofenac sodium pellets to
16 healthy volunteers in the fasted and fed states; (B) plasma diclofenac sodium
concentrationtime proles obtained following an intravenous administration of
50mg diclofenac sodium to 7 volunteers (Willis et al., 1979). Each point represents
the mean plasma diclofenac sodium concentration.
Dressman et al., 1990; Kalantzi et al., 2006), as reected by the
fed gastric dissolution medium, FeSSGF (pH 5.0) (Jantratid et al.,
2008b). At this pH, diclofenac is more soluble than in the acidic
conditions in the fasted state and part of the drug can be released.
However, the results show that the dissolution rate is not actually
faster in the fed gastric conditions. With the delayed gastric emp-
tying induced by food commonly observed in vivo, the dosage form
would reach the proximal small intestine, where signicant release
can occur, after approximately 24h (Hardy et al., 1993; Davis et al.,
1986; Coupe et al., 1991). Subsequently, along the small intestine,
as the pH increases the release is expected to continue. The disso-
lution results indicate that after approximately 6h the release of
drug in both fed and fasted states are expected to be complete.
3.4. Comparative bioavailability studies
Fig. 2A shows a slowonset of absorption after meal intake, with
signicant differences in the plasma drug concentrationtime pro-
les following oral administration of MR diclofenac sodiumpellets
in the pre- and postprandial states. Both C
max
and T
max
were sig-
nicantly different (p<0.01, Students paired t-test and Wilcoxons
Fig. 3. Mean fraction absorbed of diclofenac sodium, obtained using the numerical
deconvolution method after administration of MR diclofenac sodiumpellets (n=16)
in the fasted state () and fed state ().
signed-rank test, respectively). However, the exposure, expressed
as AUC, was not signicantly different (p>0.01, Students paired t-
test). The results are in agreement with previously reported data
(Willis et al., 1981; Riad et al., 1995).
3.5. IVIVC analysis and curve comparisons
3.5.1. Deconvolution of pharmacokinetic data
Since no intravenous data was available from this study, they
had to be taken from the literature (Fig. 2B) (Willis et al.,
1979) to perform numerical deconvolution using the WinNonlin

program. Therefore, the deconvoluted data cannot be inter-

preted as 100% accurate. By tting the intravenous plasma drug
concentrationtime data to the Exponential Decay regression
using SigmaPlot

program, tri-exponential functions were foundto

describe the prole appropriately. The unit impulse response was
then calculated from the coefcients of tri-exponential equation.
Fig. 3 shows the F
a
proles obtained following the numerical
deconvolution of the plasma proles in the fasted and fed states.
Similar patterns to those measured in the in vitro dissolution pro-
les obtained from the biorelevant methods were observed. By
contrast, the QC dissolution methods give dissolution patterns that
are obviously different from the in vivo absorption proles.
3.5.2. Curve comparisons
TheWeibull distributionhas previouslybeenappliedtotheeval-
uation of IVIVC (Nicolaides et al., 2001; Jantratid et al., 2008a). In
this study it was used to explain the incremental changes in the F
a
and F
d
curves over the same time-frame. As demonstrated in Fig. 4,
the F
d
proles from both the QC and the biorelevant methods are
steeper thantheF
a
proles inthefastedandfedstates. However, the
F
d
prole fromthe QC dissolution methods could be excluded from
IVIVC considerations since it runs much ahead of the biorelevant F
d
and the F
a
proles, and is exponential in shape. A sigmoidal shape
which can be well described by using the Weibull distribution was
observed for the F
d
proles from the Bio-Dis and the ow-through
cell methods, as well as the F
a
proles. The parameters obtained
from Weibull analysis of these proles are given in Table 4. The
d
values, which represent the overall rate of dissolution/absorption,
conrmthat thetime-frames requiredfor thedissolutioncurves are
indeed shorter than those required for the absorption curve both
in the fasted and fed states. The values, which describe the shape
of the proles, indicate that although the curve increments in vitro
440 E. Jantratid et al. / European Journal of Pharmaceutical Sciences 37 (2009) 434441
Fig. 4. Comparison of the fraction drug dissolved (F
d
) obtained fromthe biorelevant
dissolutiontests vs the fractiondrugabsorbed(Fa) inthe fastedstate (A) andfedstate
(B). The data were tted to Weibull distribution.
and in vivo are both sigmoidal (>1) the changes (particularly in
the fed state) are more obvious in vitro than in vivo.
The observed difference between the in vitro dissolution and in
vivo absorption, especially in the fed state, can be explained by the
variations in the gastric residence time and the variations of the
small intestinal pHs. The passage time used for the dissolution set-
upinthis studywas set as 2hfor thefedstatestomach. However, the
in vivo data range from2 to 4h(Hardy et al., 1993; Davis et al., 1986;
Coupe et al., 1991). It can therefore be expected that the longer fed
gastric residence time invoked for the biorelevant methods would
lead to in vitro dissolution proles closer to the in vivo absorption
prole. The other reason for the deviation of the in vitro dissolu-
tion to the in vivo absorption is that the solubility and dissolution
of diclofenac sodium is known to be inuenced strongly by pH
changes, e.g. conditions in the GI tract, owing to the weakly acidic
properties and the pK
a
of the drug itself (Chuasuwan et al., 2009).
Therefore the release of druginvivo, particularlyinthe postprandial
state, can be very sensitive to the conditions in the small intes-
tine. These, in turn, show interindividual variation as well as being
affected by the type of meal administered (Clarysse et al., 2009).
3.5.3. In vitroin vivo correlations
Fig. 5demonstrates theLevel AIVIVCplots. Inbothfastedandfed
states, the least square regressions yield essentially linear patterns
(r
2
>0.95), indicating that the proposed in vitro biorelevant disso-
Fig. 5. In vitroin vivo correlation (IVIVC) analysis representing the relationship
between the time pairs required for the same amounts of drug absorption in vivo
and drug dissolution in vitro of diclofenac sodium.
lution methods can indeed explain the absorption of diclofenac
sodium satisfactorily. The regression equations indicate that the in
vitroreleasefromBio-Dis methodranaheadof invivobyafactor of 2
in the preprandial (the regression slope =2.0459) and 3 in the post-
prandial (the regression slope =2.9977) state. For the ow-through
cell, the in vitro release was closer to the in vivo absorption in both
pre- and postprandial states (the regression slopes are 1.5888 and
1.2928, respectively).
4. Conclusions
The biorelevant dissolution testing conditions established in
this study were able to discriminate between fasted and fed state
performance of MR diclofenac sodium pellets, anticipating slower
drug release in the post- than preprandial state superior to the QC
compendial methods. Good correlations were obtained between
the time pairs of in vitro dissolution and in vivo absorption. The
results indicate that biorelevant dissolution methodology is gen-
erally appropriate for the evaluation of in vivo performance of
MR diclofenac sodium pellets, with the ow-through cell method
slightly superior to the Bio-Dis method in this example.
Acknowledgement
Part of this work has been presented at the 2007 AAPS Annual
Meeting andExposition, November 1115, 2007, SanDiegoConven-
tion Center, San Diego, CA, USA.
References
Abrahamsson, B., Johansson, D., Torstensson, A., Wingstrand, K., 1994. Evaluation of
solubilizers in the drug release testing of hydrophilic matrix extended-release
tablets of felodipine. Pharm. Res. 11, 10931097.
Amidon, G.L., Lennerns, H., Shah, V.P., Crison, J.R., 1995. A theoretical basis for a
biopharmaceutic drug classication: the correlation of in vitro drug product
dissolution and in vivo bioavailability. Pharm. Res. 12, 413420.
Balan, G., Timmins, P., Greene, D.S., Marathe, P.H., 2001. In vitro-in vivo correlation
(IVIVC) models for metforminafter administrationof modied-release(MR) oral
dosage forms to healthy human volunteers. J. Pharm. Sci. 90, 11761185.
Chuasuwan, B., Binjesoh, V., Polli, J.E., Zhang, H., Amidon, G.L., Junginger, H.E., Midha,
K.K., Shah, V.P., Stavchansky, S., Dressman, J.B., Barends, D.M., 2009. Biowaiver
monographs for immediate release solid oral dosage forms: diclofenac sodium
and diclofenac potassium. J. Pharm. Sci. 98, 12061219.
Clarysse, S., Tack, J., Lammert, F., Duchateau, G., Reppas, C., Augustijns, P., 2009. Post-
prandial evolution in composition and characteristics of human duodenal uids
in different nutritional states. J. Pharm. Sci. 98, 11771192.
E. Jantratid et al. / European Journal of Pharmaceutical Sciences 37 (2009) 434441 441
Coupe, A.J., Davis, S.S., Wilding, I.R., 1991. Variation in gastrointestinal transit
of pharmaceutical dosage forms in healthy subjects. Pharm. Res. 8, 360
364.
Davis, S.S., Hardy, J.G., Fara, J.W., 1986. Transit of pharmaceutical dosage forms
through the small intestine. Gut 27, 886892.
Dressman, J., Schamp, K., Beltz, K., Alsenz, J., 2007. Characterizing release from
lipid-based formulations. In: Hauss, D.J. (Ed.), Oral Lipid-Based Formulations:
EnhancingtheBioavailabilityof PoorlyWater-SolubleDrugs. InformaHealthcare
USA, Inc., New York, pp. 241256.
Dressman, J.B., Amidon, G.L., Reppas, C., Shah, V.P., 1998. Dissolution testing as
a prognostic tool for oral drug absorption: immediate release dosage forms.
Pharm. Res. 15, 1122.
Dressman, J.B., Berardi, R.R., Dermentzoglou, L.C., Russell, T.L., Schmaltz, S.P., Barnett,
J.L., Jarvenpaa, K.M., 1990. Upper gastrointestinal (GI) pH in young, healthy men
and women. Pharm. Res. 7, 756761.
Dressman, J.B., Reppas, C., 2000. In vitro-in vivo correlations for lipophilic, poorly
water-soluble drugs. Eur. J. Pharm. Sci. 11 (Suppl. 2), S73S80.
Emami, J., 2006. In vitroin vivo correlation: from theory to applications. J. Pharm.
Pharm. Sci. 9, 169189.
Fotaki, N., Symillides, M., Reppas, C., 2005. In vitro versus canine data for predicting
input proles of isosorbide-5-mononitrate fromoral extended release products
on a condence interval basis. Eur. J. Pharm. Sci. 24, 115122.
Frick, A., Moller, H., Wirbitzki, E., 1998. Biopharmaceutical characterization of oral
controlled/modied-release drug products. In vitro/in vivo correlation of roxa-
tidine. Eur. J. Pharm. Biopharm. 46, 313319.
Galia, E., Nicolaides, E., Hrter, D., Lbenberg, R., Reppas, C., Dressman, J.B., 1998.
Evaluation of various dissolution media for predicting in vivo performance of
class I and II drugs. Pharm. Res. 15, 698705.
Hardy, J.G., Harvey, W.J., Sparrow, R.A., Marshall, G.B., Steed, K.P., Macarios, M., Wild-
ing, I.R., 1993. Localization of drug release sites from an oral sustained-release
formulationof 5-ASA(Pentasa) inthegastrointestinal tract usinggammascintig-
raphy. J. Clin. Pharmacol. 33, 712718.
Jantratid, E., Janssen, N., Chokshi, H., Tang, K., Dressman, J.B., 2008a. Designingbiorel-
evant dissolution tests for lipid formulations: case examplelipid suspension of
RZ-50. Eur. J. Pharm. Biopharm. 69, 776785.
Jantratid, E., Janssen, N., Reppas, C., Dressman, J.B., 2008b. Dissolution media sim-
ulating conditions in the proximal human gastrointestinal tract: an update.
Pharm. Res. 25, 16631676.
Kalantzi, L., Goumas, K., Kalioras, V., Abrahamsson, B., Dressman, J.B., Reppas, C.,
2006. Characterization of the human upper gastrointestinal contents under
conditions simulating bioavailability/bioequivalence studies. Pharm. Res. 23,
165176.
Klein, S., 2005. Biorelevant dissolution test methods for modied release dosage
forms. Doctoral thesis. Department of Pharmaceutical Technology, Goethe Uni-
versity, Frankfurt am Main, 312 pp.
Klein, S., Rudolph, M.W., Skalsky, B., Petereit, H.U., Dressman, J.B., 2008. Use of
the BioDis to generate a physiologically relevant IVIVC. J. Control. Release 130,
216219.
Klein, S., Wunderlich, M., Dressman, J., 2005. Development of dissolution tests on
the basis of gastrointestinal physiology. In: Dressman, J., Krmer, J. (Eds.), Phar-
maceutical Dissolution Testing. Taylor & Francis Group, LLC, Boca Raton, FL.
Kortejrvi, H., Mikkola, J., Bckman, M., Antila, S., Marvola, M., 2002. Development of
level A, B and C in vitro-in vivo correlations for modied-release levosimendan
capsules. Int. J. Pharm. 241, 8795.
Langenbucher, F., 2005. Interpretation of in vitroin vivo time proles in terms
of extent, rate, and shape. In: Dressman, J., Krmer, J. (Eds.), Pharmaceutical
Dissolution Testing. Taylor & Francis, Florida, pp. 251279.
Levy, G., Hollister, L.E., 1964. Inter- and intrasubject variations in drug absorption
kinetics. J. Pharm. Sci. 53, 14461452.
Levy, G., Leonards, J.R., Procknal, J.A., 1965. Development of in vitro dissolution tests
which correlate quantitatively with dissolution rate-limited drug absorption in
man. J. Pharm. Sci. 51, 17191722.
Lue, B.M., Nielsen, F.S., Magnussen, T., Schou, H.M., Kristensen, K., Jacobsen, L.O.,
Mullertz, A., 2008. Using biorelevant dissolution to obtain IVIVC of solid dosage
forms containing a poorly-soluble model compound. Eur. J. Pharm. Biopharm.
69, 648657.
Nicolaides, E., Galia, E., Efthymiopoulos, C., Dressman, J.B., Reppas, C., 1999. Fore-
casting the in vivo performance of four low solubility drugs from their in vitro
dissolution data. Pharm. Res. 16, 18761882.
Nicolaides, E., Symillides, M., Dressman, J.B., Reppas, C., 2001. Biorelevant dissolution
testing to predict the plasma prole of lipophilic drugs after oral administration.
Pharm. Res. 18, 380388.
OConnor, K.M., Corrigan, O.I., 2001. Preparation and characterisation of a range of
diclofenac salts. Int. J. Pharm. 226, 163179.
Riad, L.E., Sawchuk, R.J., McAlary, M.M., Chan, K.K., 1995. Effect of food on the
multiple-peak behavior after a single oral dose of diclofenac sodium slow-
release tablet in humans. Am. J. Ther. 2, 237242.
Rossi, R.C., Dias, C.L., Donato, E.M., Martins, L.A., Bergold, A.M., Frehlich, P.E., 2007.
Development andvalidationof dissolutiontest for ritonavir soft gelatincapsules
based on in vivo data. Int. J. Pharm. 338, 119124.
Russell, T.L., Berardi, R.R., Barnett, J.L., Dermentzoglou, L.C., Jarvenpaa, K.M.,
Schmaltz, S.P., Dressman, J.B., 1993. Upper gastrointestinal pH in seventy-nine
healthy, elderly, North American men and women. Pharm. Res. 10, 187196.
Sunesen, V.H., Pedersen, B.L., Kristensen, H.G., Mllertz, A., 2005. In vivo in vitro cor-
relations for a poorly soluble drug, danazol, using the ow-through dissolution
method with biorelevant dissolution media. Eur. J. Pharm. Sci. 24, 305313.
Takka, S., Sakr, A., Goldberg, A., 2003. Development and validation of an in vitro-in
vivo correlation for buspirone hydrochloride extended release tablets. J. Control.
Release 88, 147157.
Uppoor, V.R., 2001. Regulatoryperspectives oninvitro(dissolution)/invivo(bioavail-
ability) correlations. J. Control. Release 72, 127132.
USFDA, 1997a. Guidancefor Industry: Dissolutiontestingof immediate-releasesolid
oral dosage forms. Food and Drug Administration, Center for Drug Evaluation
and Research (CDER).
US FDA, 1997b. Guidance for Industry: Extended release oral dosage forms: Devel-
opment, evaluation, and application of in vitro/in vivo correlations. Food and
Drug Administration, Center for Drug Evaluation and Research (CDER).
US FDA, 2002. Guidance for Industry: Food-effect bioavailability and fed bioequiv-
alence studies. Food and Drug Administration, Center for Drug Evaluation and
Research (CDER).
USP 30, 2008. United States Pharmacopeial Convention, Inc., Rockville, MD.
Vertzoni, M., Dressman, J., Butler, J., Hempenstall, J., Reppas, C., 2005. Simulation
of fasting gastric conditions and its importance for the in vivo dissolution of
lipophilic compounds. Eur. J. Pharm. Biopharm. 60, 413417.
Vertzoni, M., Fotaki, N., Kostewicz, E., Stippler, E., Leuner, C., Nicolaides, E., Dressman,
J., Reppas, C., 2004. Dissolution media simulating the intralumenal composi-
tion of the small intestine: physiological issues and practical aspects. J. Pharm.
Pharmacol. 56, 453462.
Wei, H., Lbenberg, R., 2006. Biorelevant dissolution media as a predictive tool for
glyburide a class II drug. Eur. J. Pharm. Sci. 29, 4552.
Willis, J.V., Kendall, M.J., Flinn, R.M., Thornhill, D.P., Welling, P.G., 1979. The pharma-
cokinetics of diclofenac sodium following intravenous and oral administration.
Eur. J. Clin. Pharmacol. 16, 405410.
Willis, J.V., Kendall, M.J., Jack, D.B., 1981. The inuence of food on the absorption of
diclofenac after single and multiple oral doses. Eur. J. Clin. Pharmacol. 19, 3337.
Wingstrand, K., Abrahamsson, B., Edgar, B., 1990. Bioavailability from felodipine
extended-release tablets with different dissolution properties. Int. J. Pharm. 60,
151156.