Journal of Controlled Release 86 (2003) 293–304

www.elsevier.com / locate / jconrel

A computational model for particle size influence on drug

absorption during controlled-release colonic delivery
Kenneth C. Waterman*, Steven C. Sutton
Pfizer Global Research and Development, Eastern Point Road, Groton, CT 06340, USA

Received 3 April 2002; accepted 1 November 2002

Abstract

The effect of particle size on the percent drug absorbed is computationally modeled for controlled-release dosage forms
that deliver drug particles to the colon. The relative benefit of reducing particle size is mapped on a diagram of the drug’s
absorption rate constant (estimated from rat intestinal perfusion, CACO-2 or human intubation permeation rates) versus the
drug’s solubility. Some drugs fall into a limit of high percentage absorption even with large particles such that particle size
reduction has little impact. Another group of drugs is solubility limited such that even with small particles, absorption is
negligible. Between the two regions, only drugs with sufficiently high absorption rates are influenced by the drug dissolution
rate and thereby the particle size. The size of this region is a function of dosing rate. Comparisons between calculated
particle size effects on colon absorption as a function of colon volume suggest caution when using animal models to predict
bioavailability from colonic drug delivery. This volume dependence also suggests that the particle size influence will vary as
a function of the digestive cycle.
 2002 Elsevier Science B.V. All rights reserved.

Keywords: Colonic drug absorption; Particle size; Controlled-release drug delivery; Drug solubilization

1. Introduction in the literature between theory and experiment [6].

The effect of particle size on crystalline drug
dissolution rates and corresponding drug absorption
has been well established both theoretically and
experimentally [1–6]. In many cases, increasing the
surface area of a crystalline drug can have a dramatic
impact on the drug absorption for low to moderate
solubility drugs. For most cases, there is agreement
*Corresponding author. Fax: 11-860-441-3972.
E-mail address: ken waterman@groton.pfizer.com (K.C. Water-
]
man).
The majority of studies establishing the role of
particle size on drug absorption have involved only
the upper gastrointestinal (GI) tract. Moreover, most
of these studies involved providing drug in an
immediate-release dosage form.
A number of drugs benefit from controlled drug
delivery. Controlled-release systems can provide
reduced frequency of dosing and can mitigate ad-
verse events by providing a reduced Cmax . In addi-
tion, by metering drug into the GI system in a slow
manner, controlled-release dosage forms can reduce
local irritation effects due to high local drug con-

0168-3659 / 02 / $ – see front matter  2002 Elsevier Science B.V. All rights reserved.
PII: S0168-3659( 02 )00418-2
294 K.C. Waterman, S.C. Sutton / Journal of Controlled Release 86 (2003) 293–304
centrations upon oral bolus administration. Control-
led-release drug delivery over longer periods (de-
pending on fed state) [7] results in a corresponding
greater fraction of the drug delivered in the colon.
Colonic drug absorption probably takes place pre-
dominantly in the ascending colon where the fluid
content remains relatively high. Although the surface
area within the colon is low compared to the small
intestinal walls, the long residence time of drug in
the ascending colon (8–34 h) [8,9], enables signifi-
cant doses of many drugs to be absorbed.
Osmotic delivery systems are practical oral, con-
trolled-release dosage forms that can deliver poorly
water-soluble drugs [10–12]. Many matrix systems
are diffusion limited (typically for high solubility
drugs), thus particle size is not a major issue. While it
is possible to design matrix systems for delivery of
poorly soluble drugs [13,14], matrix dosage forms
for these drugs are often sensitive to the GI environ-
ment since they require an erosional component to
the drug delivery. Such factors as water content,
mechanical abrasion and pH can have a dramatic
impact on the drug delivery, leading to poor in
vitro–in vivo correlations. For these reasons, such
matrix controlled-delivery systems are often less
desirable for low water-solubility drugs, especially
when significant drug delivery will occur in the
low-water environment of the colon. For osmotic
controlled-release systems, drug is delivered essen-
tially independently of the dosage form position in
the GI tract. For poorly water-soluble drugs, the
dosage form delivers a stream of particles, often
suspended in a viscous medium. These systems
therefore tend to show good in vitro–in vivo correla-
tions (see, for example, Ref. [15]). In the calculations
carried out in this study, we assume drug particles
are delivered in a steady stream as they would from
an osmotic delivery system.
When a low water-solubility drug must be ab-
sorbed in the colon due to the desired duration of
drug delivery, there are a limited number of ap-
proaches available for improving poor drug absorp-
tion. In some cases, increasing the drug dissolution
rate (decreasing particle size) may help improve drug
absorption. We were interested in estimating the
appropriate drug properties where particle size reduc-
tion could be expected to have a significant impact
on drug absorption from a controlled-release dosage
form when the drug is delivered in the colon. To
accomplish this, we developed a computational
model for calculating the fraction drug absorbed ( fa )
assuming delivery into an environment where the
amount of drug present affects the dissolution kinet-
ics of the next drug particle. Using this model, the
appropriateness of using particle size reduction for
improving lower GI drug absorption could be de-
termined a priori, thereby focusing efforts for im-
proving drug absorption appropriately.
2. Methods
The computer algorithm used in these calculations
(the details of which are described in the appendix)
was written in a modified BASIC using the
Berkeley-Madonna姠 software package [16]. A
model was developed using a graphical interface
with some equations imported from another program
[17], and others directly entered. These equations
involve modifications of previously published pro-
grams [18], which focus on release of drug in one
compartment (the colon). Drug is delivered as solid
particles, which dissolve if their concentration in the
compartment is below the drug’s solubility. This
dissolution process is controlled by the classic
Noyes–Whitney equations (as described in Refs.
[4,5]). The following assumptions are made in the
calculations: (1) spherically-shaped particles, (2) log
normal distribution of particle size fraction, and (3) a
constant diffusion layer thickness (3.5310 24 cm 2 /
min) [5]. The Runge-Kutta-4 method is used to
determine numerical solutions [19]. The model step
size was halved until the difference in results was
less than 1%. Further reduction in step size, while
improving accuracy of the predictions, unnecessarily
increases the calculation time. Parameter inputs for
each calculation are: first-order absorption rate con-
stant (k a , min 21 ), minimum volume in the compart-
ment (V, ml), zero order input rate (k 0 , mg / h), mean
particle diameter (d, mm), drug solubility (S, mg / ml),
and total absorption time. The output of each calcu-
lation is the fraction of the total dose that is absorbed
from the compartment ( fa ).
While the assumptions used in these calculations
involve simplifications, we believe the results should
 K.C. Waterman, S.C. Sutton / Journal of Controlled Release 86 (2003) 293–304
provide reasonable approximations of real systems.
As a test of the calculational model, computations
were carried out and shown to correctly estimate fa
as a function of particle size (d) for hydrocortisone
in an immediate release dosage form [5].
3. Results and discussion
In calculations of the fraction drug absorbed ( fa ),
a number of factors affect the results. The computa-
tional model used here assumes drug is delivered at a
steady rate throughout its delivery time. For these
calculations, we assume that the drug will be de-
livered in the ascending colon over a period of 6 h.
We then assume absorption in the colon will con-
tinue up to an additional 7.5 h. It should be noted
that these calculations attempt to determine only the
percent of the drug absorbed. In reality, first-pass
metabolism can significantly affect the amount of
drug in the blood stream.
In our calculations, we decided to use two repre-
sentative particle sizes. The nominal large size was
chosen to have a diameter of 100 mm, while the
small size was chosen to have a diameter of 5 mm.
The final term that significantly impacts the results
of these calculations is the volume of water into
which the particles are delivered. Whereas in the
small intestine, drug particles are swept along a
cylinder while they dissolve and absorb [20], the rate
of bulk movement in the ascending colon is suffi-
ciently slow that one can consider the system as a
static volume [21]. Water enters the colon and is
absorbed rapidly; however, the water that is effec-
tively available for drug dissolution and eventual
absorption is only the instantaneous water volume
present. In essence, the steady-state water volume
will determine the drug’s ability to dissolve. The
average volume of fluid in the ascending colon after
a meal has been measured to be 170640 ml [22].
The water content of the ascending colon is reported
to be about 90% [21]. The net result is that the
expected steady-state water volume in the ascending
colon after a meal is 150636 ml. This volume will
diminish when there is no food present. Under
fasting conditions, the volume of water present in the
ascending colon is estimated to be less than 50 ml
[23].
295
Drug absorption rates for most drugs can be
estimated using CACO-2 cell lines, single-pass
perfusion of rat intestines or intubation (in vivo)
studies [24–28]. The absorption rate, Ka , for a given
compound in the colon may be lower than that in the
small intestine due to active transporters in the small
intestine which are not expressed in the colon. For
this reason, it can be important to determine whether
active transport occurs for a given drug.
Solubility for the purpose of the computational
model used here refers to equilibrium solubility at
37 8C. We assume that particles are wetted. Obvious-
ly, the most relevant liquid to use for the solubility
measurement would be the ascending colonic fluid.
In practical terms, buffers can be used that imitate
this fluid. The pH reported in the human ascending
colon ranges considerably depending on the subject
and method of measurement. Averaging a number of
studies using radiotelemetry methodologies indicate
that the ascending colon fluid has a pH of approxi-
mately 6.1 (total of 165 subjects) with a range of
about 5.5 to 7 [29,30].
In Fig. 1 (data shown in Table 1), a map is shown
which covers a wide range of solubilities and
absorption rate constants, assuming 7.5 h of drug
Fig. 1. Particle size effects calculated for controlled release
delivery of drug to the human ascending colon. Calculations
assume 5 mg / h are delivered to the ascending colon for 6 h, with
absorption continuing for an additional 7.5 h. The regions are
defined as follows h fa is defined as the fraction of the total
delivered drug which is absorbed; If as [ fa (5 mm)2fa (100
mm)] /fa (100 mm): region A, fa (100 mm).80%; region B, fa (5
mm),20%; region C, If ,0.25; 20%,fa ,80%; region D, If .
0.25, fa (5 mm).20%j.
296 K.C. Waterman, S.C. Sutton / Journal of Controlled Release 86 (2003) 293–304

Table 1
Calculated percentage drug absorbed in the human ascending colon (volume of water5150 ml) for controlled release of 5 mg / h for 6 h
ka Solubility % Drug absorbed at 6.5 h total time % Drug absorbed at 13.5 h total time
(min 21 ) (mg / ml)
5 mm particles 100 mm particles If 5 mm particles 100 mm particles If
0.001 0.01 1.9 1.6 0.19 4.0 3.7 0.08
0.001 0.05 8.6 7.8 0.10 19.1 18.2 0.05
0.001 0.1 14.7 13.8 0.07 35.7 34.5 0.03
0.001 0.2 18.5 18.2 0.02 46.5 46.3 0
0.002 0.005 1.6 3.7
0.002 0.05 15.3 36.0
0.002 0.1 26.6 64.7
0.002 0.2 32.5 70.9
0.003 0.005 2.4 5.5
0.003 0.05 22.6 53.1
0.003 0.1 38.1 81.6
0.005 0.002 1.6 3.6
0.005 0.005 4.8 4.0 0.20 10.1 9.1 0.11
0.005 0.02 15.5 35.8
0.005 0.05 42.0 36.1 0.16 88.5 83.1 0.06
0.005 0.1 60.1 56.3 0.07 95.1 94.6 0.01
0.005 0.2 60.1 59.7 0.01 95.1 95.1 0
0.005 1.0 74.7 60.1 0.24 100.0 95.1 0.05
0.01 0.001 1.9 1.5 0.27 4.0 3.5 0.14
0.01 0.005 9.6 7.5 0.28 20.1 17.4 0.16
0.01 0.01 19.0 14.9 0.28 40.0 34.6 0.16
0.01 0.05 78.5 63.5 0.24 99.7 99.3 0
0.02 0.001 3.9 2.8 0.39 8.1 6.6 0.23
0.02 0.003 11.6 8.3 0.40 24.2 19.5 0.24
0.02 0.01 37.9 26.8 0.41 79.8 60.3 0.32
0.02 0.02 72.8 50.7 0.44 100.0 99.4 0.01
0.05 0.001 9.6 5.6 0.71 20.1 13.8 0.46
0.05 0.005 44.0 26.8 0.64 99.1 63.9 0.55
0.05 0.01 92.8 50.2 0.85 100.0 99.6 0
0.05 0.02 98.8 82.8 0.19 100.0 100.0 0
0.1 0.001 19.2 8.5 1.26 40.2 21.7 0.85
0.1 0.003 57.2 24.4 1.34 100.0 59.7 0.68
0.1 0.005 93.7 38.9 1.41 100.0 87.5 0.14
0.1 0.01 99.9 67.8 0.47 100.0 100.0 0

absorption beyond the 6 h of drug delivery in the
ascending colon (total of 13.5 h of drug absorption in
the ascending colon). In describing the drug absorp-
tion behavior, we have defined two terms: fa is
defined as the fraction of the total dose delivered that
is absorbed, and the improvement factor, If , is
defined as [ fa (5 mm)2fa (100 mm)] /fa (100 mm). In
this map, we have defined four regions corre-
sponding to different behaviors.
3.1. Region A, fa (100 m m).80%
The drug absorption is high, fa .80%, with the
large, 100-mm diameter particles. In this region, the
absorption is sufficiently high that any benefit
achieved by reducing particle size is anticipated to be
unmeasurable. This region corresponds to drugs with
relatively high solubility (in the colonic pH range of
5–7.5) and high permeability. This situation falls
loosely into the biopharmaceutics classification sys-
tem (BCS) proposed for immediate release dosage
forms by Amidon and co-workers [31,32] as Class I
(high solubility, high permeability). Since the drug is
delivered over an extended period of time, the
definition of high solubility could be broadened to
include an equilibrium solubility of greater than
 K.C. Waterman, S.C. Sutton / Journal of Controlled Release 86 (2003) 293–304
about 0.1 mg / ml. The map in Fig. 1 shows that this
and other regions are not strictly quadrants, but
rather irregularly shaped areas.
3.2. Region B, fa (5 m m),20%
The drug absorption is low even with the small,
5-mm diameter particles ( fa ,20%). In this region,
even if particle size reduction has a significant
relative impact on drug absorption, the absolute
absorption is sufficiently low that improvements with
particle size reduction will not be significant. Drugs
in this region are solubility, rather than dissolution
rate, limited. In addition, the absorption rate is not
sufficiently fast to deplete the saturated drug in the
colon. These drugs fall broadly into Class IV of the
BCS (low solubility, low permeability), though
again, an irregular area is defined.
In the remaining regions, the relative effect of
particle size is evaluated. This effect is classified by
the percentage enhancement seen when going from
the large to the small particles. For example, if a
10% absorption with large particles becomes 50%
absorption with the small particles, we would consi-
der the impact to be a 400% improvement.
3.3. Region C, If ,0.25; 20%, fa ,80%
The impact of particle size is negligible to moder-
ate on the percentage drug absorbed (,25% relative
improvement in fa as the average particle diameter is
reduced from a 100 to a 5 mm diameter). This region
can be characterized as being absorption rate limited
such that particle size reduction does not affect the
percentage drug absorbed. The drugs in this group
fall broadly into Class III of the BCS (high solu-
bility, low permeability). The definition of high
solubility for a controlled-release dosage form could
be defined to include solubilities considerably lower
than for immediate-release dosage forms, due partial-
ly to the metering of the drug into the colon by
controlling its release from the dosage form.
3.4. Region D, If .0.25, fa (5 m m).20%
The effect of particle size reduction is predicted to
be moderate to significant (.25% relative improve-
ment in fa as particle diameter is reduced from 100 to
297
5 mm diameter). In this region, drug dissolution is
rate limiting. Drugs in this region fall broadly into
Class II of the BCS (low solubility, high permeabili-
ty). This region is where size reduction of particles is
likely to be worthwhile. Every drug is unique,
however, and the ability to predict the benefit by a
priori methods is limited. Still, the likelihood of
improving bioavailability for a drug falling in other
regions (A–C) is less than for a drug falling in
region D.
If one assumes absorption occurs over a shorter
time period, the regions shown in Fig. 1 shift
considerably. Fig. 2 shows a mapping of particle size
effects for absorption as shown in Fig. 1 with the
absorption assumed to stop 30 min beyond the 6 h of
delivery. This situation will mimic a long duration
controlled-release dosage form such that the resi-
dence time in the ascending colon is comparable to
the dosing time. As can be seen, the region of
maximum particle size impact shifts towards higher
solubilities. Overall, as one would expect, the ab-
sorption percentages are reduced. As an example, if
the solubility is assumed to be 0.02 mg / ml and k a to
be 0.02 min 21 under the two previously stated
conditions, reducing the particle size results in a
significant improvement in f a when absorption oc-
curs over a time similar to the delivery duration (this
point lies within region D, see Table 1, Fig. 2).
However, if absorption is allowed to continue a
Fig. 2. Particle size effects calculated for controlled-release
delivery of drug to the human ascending colon (assuming a
volume of water of 150 ml). Calculations assume 5 mg / h are
delivered to the colon for 6 h, with absorption continuing for an
additional 0.5 h. The regions are defined as in Fig. 1.
298 K.C. Waterman, S.C. Sutton / Journal of Controlled Release 86 (2003) 293–304
further 7.5 h, absorption of the larger particles
completes such that the benefit of particle size
reduction becomes inconsequential (the point now
falls in region A in Fig. 1).
One consequence of delivering a drug slowly in a
controlled-release dosage form rather than in an
immediate release bolus is that there is effectively
more dilution of the drug in the former case. The
result is that solubilities that would be considered
low for immediate release dosage forms (e.g., 0.1
mg / ml), are effectively high when released slowly in
controlled-release systems.
It is interesting to compare the colonic absorption
predictions for controlled-release dosage forms in the
human to those in the dog, where many such systems
are tested [33]. The effective volume of water in the
dog colon is difficult to estimate, and has not been
reported in the literature, but is probably around 50
ml. Using this estimate with the same calculational
procedure as used above (i.e., 5 mg / h dosing for 6-h
followed by an additional 7.5 h of absorption), a new
plot can be generated as shown in Fig. 3 (data in
Table 2). In comparing Figs. 1 and 3, it becomes
obvious that a straightforward use of a dog for
estimating drug bioavailability for controlled-release
dosage forms could grossly underestimate not only
the extent of absorption, but also the relative benefit
Fig. 3. Particle size effects calculated for controlled-release
delivery of drug to the dog colon or the human colon under fasting
conditions (assuming a volume of water of 50 ml). Calculations
assume 5 mg / h are delivered to the colon for 6 h, with absorption
continuing for an additional 7.5 h. The regions are defined as in
Fig. 1.
of a particle size reduction. In particular, the region
of benefit for particle size reduction (region D)
decreases with the reduced volume of water in the
dog colon. This situation can be improved by
reducing the dosing rate for the dog. For example, in
Fig. 4 (data in Table 3), a dosing rate of 1 mg / h is
used with the dog model (i.e., 50 ml of water in the
colon). As can be seen in comparing Figs. 1 and 4,
region D (where particle size reduction is expected to
have a significant impact on bioavailability) now
covers a similar space. The drug concentration in the
human colon and each of the dog colon models after
1 h assuming no drug absorption corresponds to
0.033 mg / ml (5 mg / 150 ml, for the human), 0.100
mg / ml (5 mg / 50 ml, for the dog) and 0.020 mg / ml
(1 mg / 50 ml, for the dog assuming a lower dosing
rate). As these concentrations show, delivering drug
to a dog at 1 mg / h more closely mimics dosing a
human at 5 mg / h (0.033 vs. 0.020 mg / ml) than does
dosing the dog at the same rate (0.100 mg / ml). In
summary, a dosing rate in the dog that takes into
account the relative colonic water content is expected
to make the dog a better model for solubility
dependent absorption in the lower GI tract.
Under fasting conditions, the human ascending
colon water volume has been suggested to be less
than 50 ml [23], about the same as that used in the
calculation for the fed dog. Based on these calcula-
tions, therefore, we would predict that the drug
properties where particle size impacts bioavailability
in the colon will be narrower and somewhat different
from those of a fed person.
If one delivers the drug to the human ascending
colon at a reduced rate, the impact of particle size
reduction is greater and the region of impact (region
D) is greater. This can be seen in Fig. 5 (data in
Table 4). In comparing Figs. 1 and 5, region D for
the lower dosing rate encompasses lower absorption
rate constants and greater drug solubilities. In con-
trast, Fig. 6 (data in Table 5) shows that at a high
dosing rate, there is only a very narrow region where
particle size is predicted to have a significant impact.
It is interesting to see where real drugs fall in
these categories based on their solubilities and
permeabilities. Towards this end, the values and
predicted behaviors of 10 randomly selected drugs
are shown in Table 6. As can be seen, there are drugs
that fall into each category at the low doses. At
 K.C. Waterman, S.C. Sutton / Journal of Controlled Release 86 (2003) 293–304 299

Table 2
Calculated percentage drug absorbed in the dog colon or human fasting colon (volume of water550 ml) for controlled release of 5 mg / h for
6h
ka Solubility % Drug absorbed at 6.5 h total time % Drug absorbed at 13.5 h total time
(min 21 ) (mg / ml)
5 mm particles 100 mm particles If 5 mm particles 100 mm particles If
0.001 0.01 0.6 0.6 0 1.3 1.3 0
0.001 0.05 3.1 2.9 0.07 6.6 6.4 0.03
0.001 0.5 18.5 18.3 0.01 46.5 46.3 0
0.001 1.0 18.5 18.5 0 46.5 46.4 0
0.005 0.05 15.6 14.3 0.09 33.1 31.6 0.05
0.005 0.01 3.2 2.9 0.10 6.7 6.4 0.05
0.005 0.1 29.7 27.8 0.07 64.7 62.2 0.04
0.005 0.5 60.1 60.0 0 95.1 95.1 0
0.005 1.0 64.4 60.1 0.07 100.0 95.1 0.05
0.01 0.001 0.7 0.6 0.17 1.4 1.3 0.08
0.01 0.005 3.2 2.9 0.10 6.7 6.3 0.06
0.01 0.01 6.4 5.7 0.12 13.4 12.6 0.06
0.01 0.05 31.1 27.8 0.12 66.1 61.8 0.07
0.01 0.1 58.5 52.8 0.11 98.9 98.2 0.01
0.05 0.001 3.2 2.5 0.28 6.7 5.7 0.18
0.05 0.005 16.1 12.3 0.31 33.6 28.2 0.19
0.05 0.01 32.1 24.3 0.32 67.1 55.3 0.21
0.05 0.05 98.8 93.5 0.06 100.0 100.0 0
0.1 0.001 6.5 4.3 0.51 13.5 10.2 0.32
0.1 0.005 32.2 21.1 0.53 67.2 49.3 0.36
0.1 0.01 64.0 40.6 0.58 100.0 90.0 0.11
0.1 0.05 99.9 99.5 0 100.0 100.0 0

higher doses, one predicts that few drugs will benefit

from particle size reduction.

4. Conclusions

Since drug delivery in controlled-release dosage

Fig. 4. Particle size effects calculated for controlled-release
delivery of drug to the dog colon or the fasting human colon
(assuming a volume of water of 50 ml). Calculations assume 1
mg / h are delivered to the colon for 6 h, with absorption
continuing for an additional 7.5 h. The regions are defined as in
Fig. 1.
forms for longer durations involves drug release into
the colon, determining factors that affect the drug
absorption in the colon can be important in the
design of the dosage form. Colonic drug absorption
is sometimes poor due to limited drug solubility. A
computational model for controlled-release drug
dissolution and absorption was used to examine the
effect of particle size reduction on improving bio-
availability. It was found that particle size will only
have a significant impact on the drug absorption
(bioavailability) for a limited class of drugs roughly
falling into the BCS II classification (low solubility,
high permeability). Four regions were defined based
on the drug solubility, absorption rate constant,
dosing rate and dosing duration. These regions
300 K.C. Waterman, S.C. Sutton / Journal of Controlled Release 86 (2003) 293–304

Table 3
Calculated percentage drug absorbed in the dog colon or fasting human colon (volume of water550 ml) for controlled release of 1 mg / h for
6h
ka Solubility % Drug absorbed at 6.5 h total time % Drug absorbed at 13.5 h total time
(min 21 ) (mg / ml)
5 mm particles 100 mm particles If 5 mm particles 100 mm particles If
0.001 0.001 0.3 0.3 0 0.7 0.6 0.17
0.001 0.01 3.1 2.6 0.19 6.6 6.0 0.10
0.001 0.1 18.5 17.3 0.07 46.4 45.5 0.02
0.001 0.5 18.5 18.5 0 46.5 46.4 0
0.001 1.0 18.5 18.5 0 46.5 46.5 0
0.0025 0.025 18.2 15.1 0.21 40.1 36.3 0.10
0.0025 0.05 31.6 27.3 0.16 72.3 67.2 0.08
0.0025 0.1 38.8 37.4 0.04 78.6 78.1 0.01
0.005 0.005 7.9 6.1 0.30 16.7 14.4 0.16
0.005 0.01 15.6 12.0 0.30 33.1 28.6 0.16
0.005 0.05 58.8 49.0 0.20 95.0 93.2 0.02
0.005 1.0 76.8 60.1 0.28 100.0 95.1 0.05
0.01 0.001 3.2 2.3 0.39 6.7 5.5 0.22
0.01 0.005 15.9 11.3 0.41 33.4 27.1 0.23
0.01 0.01 31.0 22.1 0.40 66.0 60.7 0.09
0.01 0.05 80.0 74.6 0.07 99.7 99.6 0
0.01 0.1 80.0 79.1 0.01 99.7 99.7 0
0.01 1.0 100.0 80.0 0.25 100.0 99.7 0
0.025 0.0025 20.0 11.8 0.69 41.8 29.2 0.43
0.025 0.01 76.3 43.6 0.75 100.0 95.6 0.05
0.025 0.025 94.8 83.1 0.14 100.0 100.0 0
0.05 0.001 16.0 7.5 1.13 33.5 19.3 0.74
0.05 0.005 78.0 35.0 1.23 100.0 81.7 0.22
0.05 0.01 98.8 62.2 0.59 100.0 100.0 0
0.075 0.005 99.6 42.0 1.37 100.0 91.8 0.09
0.1 0.001 31.7 10.6 1.99 66.6 27.9 1.39
0.1 0.01 99.9 75.8 0.32 100.0 100.0 0
0.1 0.1 99.9 99.8 0 100.0 100.0 0
0.1 1.0 100.0 99.9 0 100.0 100.0 0
0.5 0.001 100.0 15.9 5.29 100.0 42.3 1.36

correspond to drug absorption being sufficiently high
or low as to not be changed by particle size, a region
where the particles size has little impact since
dissolution is not rate limiting, and finally a region
where particle size has a significant impact. These
regions depend heavily on the volume of water
available for drug dissolution; consequently, using
such animal models as dogs can provide non-predic-
tive bioavailability estimates due to the volume
differences. Even the relative benefit of particle size
reduction can be completely masked in the reduced
volume of the canine colon. It was found that
proportional dose reduction could help make the dog
a better model for predicting the efficacy of particle
size reduction for improving colonic bioavailability
in humans. It is also suggested that the fed state of
the subject can impact the effectiveness of particle
size reduction on colonic absorption by varying the
water volume in the colon.
Acknowledgements
We would like to thank Esther Ladipo for her help
in running the computer simulations used in this
project. We would also like to thank Dr. Blair West
of Bend Research Institute for his assistance in
tracking down estimates of the human ascending
colon water volume and colon pH, and Dr. Kevin
Johnson (of Intellipharm, LLC) for help in develop-
ment of earlier versions of this simulation program.
 K.C. Waterman, S.C. Sutton / Journal of Controlled Release 86 (2003) 293–304 301

Fig. 5. Particle size effects calculated for controlled-release
delivery of drug to the human colon (assuming a volume of water
of 150 ml). Calculations assume 1 mg / h are delivered to the colon
for 6 h, with absorption continuing for an additional 7.5 h. The
regions are defined as in Fig. 1.
Fig. 6. Particle size effects calculated for controlled-release
delivery of drug to the human colon (assuming a volume of water
of 150 ml). Calculations assume 25 mg / h are delivered to the
colon for 6 h, with absorption continuing for an additional 7.5 h.
The regions are defined as in Fig. 1.

Table 4
Calculated percentage drug absorbed in the human ascending colon (volume of water5150 ml) for controlled release of 1 mg / h for 6 h
ka Solubility % Drug absorbed at 6.5 h total time % Drug absorbed at 13.5 h total time
(min 21 ) (mg / ml)
5 mm particles 100 mm particles If 5 mm particles 100 mm particles If
0.001 0.01 8.6 6.0 0.43 19.1 16.2 0.18
0.001 0.05 18.5 17.2 0.08 46.5 45.6 0.02
0.001 0.1 18.5 18.2 0.02 46.5 46.2 0.01
0.001 0.5 18.5 18.5 0 46.5 46.4 0
0.001 1.0 18.5 18.5 0 46.5 46.5 0
0.002 0.01 17.1 11.6 0.47 38.1 31.1 0.23
0.002 0.02 28.6 20.7 0.38 66.9 56.9 0.18
0.002 0.05 32.9 30.9 0.06 71.0 70.2 0.03
0.002 0.1 32.9 32.4 0.02 71.0 70.8 0
0.002 0.2 32.9 32.7 0.01 71.0 71.0 0
0.005 0.002 9.5 5.6 0.70 20.0 14.7 0.36
0.005 0.005 22.8 13.5 0.69 49.0 35.9 0.36
0.005 0.01 41.9 25.5 0.64 88.5 66.9 0.32
0.005 0.02 60.1 43.4 0.38 95.1 91.8 0.04
0.01 0.001 9.5 4.9 0.94 20.0 13.0 0.54
0.01 0.002 18.9 9.6 0.97 39.8 25.7 0.55
0.01 0.005 45.1 23.0 0.96 94.4 59.9 0.58
0.01 0.01 78.1 42.0 0.86 99.7 94.8 0.05
0.01 0.02 80.0 65.6 0.22 99.7 99.5 0
0.02 0.001 19.0 7.7 1.47 39.9 20.9 0.91
0.02 0.002 37.5 15.1 1.48 79.4 40.1 0.98
0.02 0.005 86.7 34.9 1.48 100.0 84.3 0.19
0.02 0.01 92.4 60.1 0.54 100.0 99.9 0
0.1 0.002 99.9 27.1 2.69 100.0 67.4 0.48
0.1 0.003 99.9 38.3 1.61 100.0 86.1 0.16
302 K.C. Waterman, S.C. Sutton / Journal of Controlled Release 86 (2003) 293–304

Table 5
Calculated percentage drug absorbed in the human ascending colon (volume of water5150 ml) for controlled release of 25 mg / h for 6 h
ka Solubility % Drug absorbed at 6.5 h total time % Drug absorbed at 13.5 h total time
(min 21 ) (mg / ml)
5 mm particles 100 mm particles If 5 mm particles 100 mm particles If
0.001 0.005 0.2 0.2 0 0.4 0.4 0
0.001 0.01 0.4 0.4 0 0.8 0.8 0
0.001 0.05 1.9 1.8 0.06 4.0 3.9 0.03
0.001 0.1 3.7 3.6 0.03 7.9 7.8 0.01
0.001 0.5 14.7 14.5 0.01 35.7 35.5 0.01
0.001 1.0 18.5 18.5 0 46.5 46.4 0
0.005 0.1 18.5 17.6 0.05 39.5 38.4 0.03
0.005 0.5 60.1 58.6 0.03 95.1 95.1 0
0.005 1.0 60.1 60.1 0 95.1 95.1 0
0.01 0.001 0.4 0.4 0 0.8 0.8 0
0.01 0.002 0.8 0.7 0.14 1.6 1.5 0.07
0.01 0.05 19.0 17.5 0.09 40.0 38.3 0.04
0.01 0.1 36.9 34.4 0.07 78.9 75.4 0.05
0.01 0.5 80.0 79.9 0 99.7 99.7 0
0.05 0.005 9.7 8.1 0.20 20.2 18.0 0.12
0.05 0.01 19.4 16.0 0.21 40.4 35.7 0.13
0.05 0.05 93.5 73.5 0.27 100.0 100.0 0
0.05 0.1 98.8 98.2 0.01 100.0 100.0 0
0.1 0.001 3.9 2.9 0.34 8.1 6.7 0.21
0.1 0.002 7.8 5.8 0.34 16.2 13.4 0.21
0.1 0.005 19.4 14.4 0.35 40.4 33.1 0.22
0.1 0.01 38.7 28.4 0.36 80.7 64.4 0.25
0.1 0.05 99.9 98.4 0.02 100.0 100.0 0
0.1 0.1 99.9 99.8 0 100.0 100.0 0

Appendix A
The equations and methods utilized in this model
were modified from the work of Johnson and co-
workers [4,5]. Drug delivery was defined as a zero
order input, k 0 (mg / h), consisting of the total dose
(mg) and the duration of the zero order release (t end ).
No lagtime was specified (t start ), hence the drug
release began immediately; that is time, t505t start .
The drug release then continued until t5t end (unless
t end ,t e , see below). Drug absorption continued until
the specified user input t e , defined as that time when

Table 6
Predicted particle size effects on drug absorption for 10 randomly selected drugs (from Ref. [24])
Drug ka Solubility Particle size effect prediction
No. (min 21 ) (mg / ml)
5 mg / h 5 mg / h 1 mg / h 1 mg / h 25 mg / h
(150 ml volume) (50 ml volume) (50 ml volume) (150 ml volume) (150 ml volume)
I 0.009 0.100 A A A A C
II 0.041 0.002 D B D D B
III 0.014 0.014 C B C A C
IV 0.003 39 A A A A A
V 0.031 58 A A A A A
VI 0.031 0.055 A A A A C
VII 0.009 0.01 C B B A C
VIII 0.008 0.0006 B B B B B
IX 0.062 0.002 D B D D B
X 0.034 0.004 D B D A C
Letters correspond to regions as described in Fig. 1.
 K.C. Waterman, S.C. Sutton / Journal of Controlled Release 86 (2003) 293–304
excretion occurred. Although defined in terms of
excretion, t e can also be defined as that time when
peristalsis carried all drug (undelivered, dissolved
and solid) into a region of the colon where there was
negligible dissolution and / or absorption.
The particle size distribution of delivered drug was
assumed to be log-normally distributed. The geomet-
ric mean radius ( m ) was calculated from the input
mean particle size diameter, while the standard
deviation (s ) was assumed to be 2.4 mm. Eight
equal, logarithmically spaced particle size ‘‘bins’’
were calculated between the upper (r min 5 mg s 3g ) and
lower radii limits (r max 5 mg /s 3g ). The fraction of
total mass distributed into each ‘‘bin’’ was simulated
through the use of the log-normal probability density
function [5]:
f(r 0 ) 5 ]]]
1
ΠS F log r 0 2 log mg
] exp 2 0.5 ]]]]]
log sg 2p log sg GD2
(A.1)
This became the initial particle size distribution.
Dissolution was simulated using a previously
reported [5] Noyes–Whitney-type equation:
dXsi DS
]]
dt
F Xd
5 ] ? Cs 2 ]
h V
G (A.2)
where Xsi is the mass of solid drug at any time, D is
the drug diffusion coefficient 3.52310 24 cm 2 / mg
[5], Cs is the drug solubility (a user input), h is the
diffusion layer thickness (lesser value of the particle
radius and 30 mm), Xd is the mass of dissolved drug
at any time, and V is the colon volume (a user input).
As described in Ref. [5], the surface area term of the
Noyes–Whitney equation, S, was calculated for
spherically-shaped particles.
The mass of drug in the colon at any time, t, was
the sum of the amount remaining in the colon at the
end of the last step, [Xd (t21)1Xs (t21)], and the
amount delivered by the hypothetical controlled-
release formulation, k 0 (t), minus the amount ab-
sorbed, k *a Xd (t). The amount of drug absorbed, fa ,
was calculated from:
k a Xd
]] (A.3)
k0
303
References
[1] J.H. Fincher, Particle size of drugs and its relationship to
absorption and activity, J. Pharm. Sci. 57 (1968) 1825–1835.
[2] N. Kaneniwa, N. Wataki, Dissolution of slightly soluble
drugs. IV. Effect of particle size of sulfonamides on in vitro
dissolution rate and in vivo absorption rate, and their relation
to solubility, Chem. Pharm. Bull. 26 (1978) 813–826.
[3] N. Kaneniwa, N. Watari, H. Ijima, Dissolution of slightly
soluble drugs. V. Effect of particle size on gastrointestinal
drug absorption and its relation to solubility, Chem. Pharm.
Bull. 26 (1978) 2603–2614.
[4] R.J. Hintz, K.C. Johnson, The effect of particle size dis-
tribution on dissolution rate and oral absorption, Int. J.
Pharm. 51 (1989) 9–17.
[5] A.T.K. Lu, M.E. Frisella, K.C. Johnson, Dissolution model-
ing: factors affecting the dissolution rates of polydisperse
powders, Pharm. Res. 10 (1993) 1308–1314.
[6] A. Scholz, B. Abrahamsson, S.M. Diebold, E. Kostewicz,
B.I. Polentarutti, A.-L. Ungell, J.B. Dressman, Influence of
hydrodynamics and particle size on the absorption of
felodipine in labradors, Pharm. Res. 19 (2002) 42–96.
[7] W.F. Ganog, in: Review of Medical Physiology, 13th
Edition, Appleton and Lange, Norwalk, CT, 1987, p. 424.
[8] M. Bouchoucha, S.R. Thomas, Error analysis of classic
colonic transit time estimates, Am. J. Gastrointest. Liver
Physiol. 279 (2000) G520–G527.
[9] J. Hammer, S.F. Phillips, Fluid loading of the human colon:
effects on segmental transit and stool composition, Gastroen-
terology 105 (1993) 988–998.
[10] R.K. Verma, D.M. Krishna, S. Garg, Formulation aspects in
the development of osmotically controlled oral drug delivery
systems, J. Control. Release 79 (2002) 7–27.
[11] R.K. Verma, B. Mishra, S. Garg, Osmotically controlled oral
drug delivery, Drug Dev. Ind. Pharm. 26 (2000) 695–708.
[12] F. Theeuwes, Elementary osmostic pump, J. Pharm. Sci. 64
(1975) 1987–1991.
[13] P. Colombo, R. Bettini, P. Santi, N.A. Peppas, Swellable
matrices for controlled drug delivery: gel-layer behaviour,
mechanism and optimal performance, PSTT 3 (2000) 198–
204.
[14] K.V.R. Rao, K.P. Devi, Swelling controlled-release systems:
recent developments and applications, Int. J. Pharm. 48
(1988) 1–13.
[15] M. Pitsiu, G. Sathyan, S. Gupta, D. Verotta, A sermiparamet-
ric deconvolution model to establish in vivo–vitro correla-
tion applied to OROS oxybutynin, J. Pharm. Sci. 90 (2001)
702–712.
[16] R.I. Macey, G.F. Oster (programmed by T. Zahnley),
Berkeley Madonna姠 8.0.1, available from www.ber-
keleymadonna.com, Berkeley, CA, 2001.
[17] Systems Thinking Experiential Learning Laboratory (STEL-
LA), software available from High Performance Systems,
Inc., Lebannon, NH, 2000.
[18] S.C. Sutton, The use of pharmacokinetic profiles to drive the
formulation development of modified release dosage forms,
304 K.C. Waterman, S.C. Sutton / Journal of Controlled Release 86 (2003) 293–304
in: AAPS Conference on Pharmaceutics and Drug Delivery,
AAPS, Arlington, VA, 2002.
[19] F. Bornemann, Runge-Kutta methods, trees and mathe-
matica, SIAM J. Appl. Math. 2 (2001) 1–13.
[20] D.-M. Oh, R.L. Curl, G.L. Amidon, Estimating the fraction
dose absorbed from suspensions of poorly soluble com-
pounds in humans: a mathematical model, Pharm. Res. 10
(1993) 264–270.
[21] J.S. Scolapio, M. Camiller, M.R. von der Ohe, R.B. Hanson,
Ascending colon response to feeding: evidence for a 5-
hydroxytryptamine-3 mechanism, Scand. J. Gastroenterol. 30
(1995) 562–567.
[22] A.D. Badley, M. Camilleri, M.K. O’Connor, Noninvasive
measurement of human ascending colon volume, Nucl. Med.
Commun. 14 (1993) 485–489.
[23] C. Wilson, personal communication.
[24] S.C. Sutton, M.T.S. Rinaldi, J.M. McCarthy, K.E. Vuk-
ovinsky, A statistical method for the determination of
absorption rate constant estimated using the single pass
intestinal perfusion model and multiple linear regression, J.
Pharm. Sci. 91 (2002) 1046–1053.
[25] H. Lennernas, Animal perfusion studies, Drugs Pharm. Sci.
106 (2000) 73–98.
[26] S.C. Sutton, M.T.S. Rinaldi, K.E. Vukovinsky, Comparison
of the gravimetric, phenol red and 14 C-PEG-3350 methods to
determine water absorption in the rat single pass intestinal
perfusion model, Pharm. Sci. 3(3) (2001) article 25.
[27] H. Bohets, P. Annaert, G. Mannens, L. Van Beijsterveldt, K.
Anciaux, P. Verboven, W. Meuldermans, K. Lavrijsen, Strate-
gies for absorption screening in drug discovery and develop-
ment, Curr. Top. Med. Chem. 1 (2001) 367–383.
[28] H. Lennernas,¨ K. Palm, U. Fagerholm, P. Artursson, Com-
parison between active and passive transport in human
intestinal epithelial (Caco-2) cells in vitro and human
jejunum in vivo, Int. J. Pharm. 127 (1996) 103–107.
[29] S.G. Nugent, D. Kumar, D.S. Rampton, D.F. Evans, Intesti-
nal luminal pH in inflammatory bowel disease: possible
determinants and implications for therapy with amino-
salicylates and other drugs, Gut 48 (2001) 571–577.
[30] P. Gruber, M.A. Longer, J.R. Robinson, Some biological
issues in oral, controlled drug delivery, Adv. Drug Deliv.
Rev. 1 (1987) 1–18.
[31] G.L. Amidon, H. Lennernas, ¨ V.P. Shah, J.R. Crison, A
theoretical basis for a biopharmaceutic drug classification:
the correlation of in vitro and in vivo bioavailability, Pharm.
Res. 12 (1995) 413–420.
[32] E. Lipka, G.L. Amidon, Setting bioequivalence requirements
for drug development based on preclinical data: optimizing
oral drug delivery systems, J. Control. Release 62 (1999)
41–49.
[33] S.C. Sutton, Use of the dog colon model and a physiologic–
pharmacokinetic computer model to simulate controlled
release plasma profiles, Proc. Inter. Symp. Control. Release
Bioactive Mater. 27 (2000) 57–58.