Extravascular transport of drugs in tumor tissue: Effect of lipophilicity on diffusion of tirapazamine analogues in multicellular layer cultures

Article abstract of DOI:10.1021/jm049549p

The extravascular diffusion of antitumor agents is a key determinant of their therapeutic activity, but the relationships between physicochemical properties of drugs and their extravascular transport are poorly understood. It is well-known that drug lipop

Full text of DOI:10.1021/jm049549p

J. Med. Chem. 2005, 48, 1079-1087
1079
Extravascular Transport of Drugs in Tumor Tissue: Effect of Lipophilicity on
Diffusion of Tirapazamine Analogues in Multicellular Layer Cultures
Frederik B. Pruijn,* Joanna R. Sturman, H. D. Sarath Liyanage, Kevin O. Hicks, Michael P. Hay, and
William R. Wilson
Auckland Cancer Society Research Centre, Faculty of Medical and Health Sciences, The University of Auckland,
Private Bag 92019, Auckland, New Zealand
Received June 8, 2004
The extravascular diffusion of antitumor agents is a key determinant of their therapeutic
activity, but the relationships between physicochemical properties of drugs and their extravas-
cular transport are poorly understood. It is well-known that drug lipophilicity plays an
important role in transport across biological membranes, but the net effect of lipophilicity on
transport through multiple layers of tumor cells is less clear. This study examines the influence
of lipophilicity (measured as the octanol-water partition coefficient P) on the extravascular
transport properties of the hypoxic cytotoxin tirapazamine (TPZ, 1) and a series of 13 neutral
analogues, using multicellular layers (MCLs) of HT29 human colon carcinoma cells as an in
vitro model for the extravascular compartment of tumors. Flux of drugs across MCLs was
determined using diffusion chambers, with the concentration-time profile on both sides of the
MCL measured by HPLC. Diffusion coefficients in the MCLs (DMCL) were inversely proportional
to Mr0.5 (Mr, relative molecular weight), although this was a minor contributor to differences
between compounds over the narrow Mr range investigated. Differences in lipophilicity had a
larger effect, with a sigmoidal dependence of DMCL on log P. Correcting for Mr differences,
lipophilic compounds (log P > 1.5) had ca. 15-fold higher DMCL than hydrophilic compounds
(log P < -1). Using a pharmacokinetic/pharmacodynamic (PK/PD) model in which diffusion in
the extravascular compartment of tumors is considered explicitly, we demonstrated that hypoxic
cell kill is very sensitive to changes in extravascular diffusion coefficient of TPZ analogues
within this range. This study shows that simple monosubstitution of TPZ can alter log P enough
to markedly improve extravascular transport and activity against target cells, especially if
rates of metabolic activation are also optimized.
Introduction
knowledge of the relationship between drug physico-
chemical properties and transport parameters) is lack-
ing.
To reach their target cells in solid tumors, most
antitumor agents must diffuse through the extravas-
cular compartment. The importance of this problem is
well-recognized for high Mr (relative molecular weight)
therapeutic agents such as antibodies and other complex
biological molecules;1,2 less is understood about ex-
travascular transport of small molecule drugs, although
it has been suspected for decades that this is a signifi-
cant limitation in the chemotherapy of solid tumors.3-9
This problem is expected to be particularly important
for hypoxic cytotoxins such as tirapazamine (TPZ; 1,2,4-
benzotriazin-3-amine 1,4-dioxide, 1)10,11 since these
must diffuse relatively large distances from functional
blood vessels to reach their target (hypoxic) cells. In
addition, TPZ, like most other hypoxic cytotoxins, is
activated by metabolism under hypoxia12,13 and will
necessarily be consumed as it diffuses into hypoxic
zones. Studies with hypoxic 3D cell cultures have
confirmed that TPZ is metabolized rapidly enough to
impede its penetration into hypoxic zones.14-18 It may
be feasible to improve the diffusion range of TPZ by
designing analogues with higher diffusion coefficients
in tissue, but the required theoretical framework (i.e.,
The available evidence suggests that TPZ, like many
other anticancer drugs, enters cells by passive diffu-
sion.13,18 Given the well-established relationship be-
tween lipophilicity (typically measured by the octanol-
water partition coefficient P) and plasma membrane
permeability for small molecules,19,20 it is a reasonable
hypothesis that lipophilic drugs will be able to access
the transcellular transport route (through cells), while
more hydrophilic drugs will be confined to the paracel-
lular route (between cells). Depending on the interstitial
volume fraction, tortuosity of the extracellular path, and
matrix composition, restriction to the paracellular route
has the potential to greatly slow transport. Therefore,
it is reasonable to expect P to be a determinant of
extravascular transport in tumors. However, to our
knowledge, there has not been any direct or quantitative
determination of the effect of lipophilicity on the ability
of anticancer drugs (or other molecules) to diffuse the
large intercapillary distances typical of tumors.
We and others have developed a novel experimental
model for measuring the extravascular diffusion coef-
ficients of anticancer drugs (Figure 1).15,21,22 In this
model, tumor cells are grown on permeable collagen-
coated Teflon support membranes to form a multicel-
lular layer (MCL) of up to 20 cell layers in thickness at
* To whom corespondence should be addressed. Phone: 64-9-
3737599,
extension
86939.
Fax:
64-9-3737571.
E-mail:
f.pruijn@auckland.ac.nz.
10.1021/jm049549p CCC: $30.25 © 2005 American Chemical Society
Published on Web 01/27/2005

1080 Journal of Medicinal Chemistry, 2005, Vol. 48, No. 4
Pruijn et al.
Scheme 1a
a Reagents: (i) concentrated HCl, aniline, DME; (ii) 15 + 18, DMSO; (iii) H2, Pd/C, EtOH; (iv) MCPBA, DCM; (v) CF3CO3H, CHCl3.
uted parameter PK/PD model to examine the effect of
changing the DMCL of TPZ on hypoxic cell killing in
tumor tissue.
Chemistry
TPZ (1) and the analogues 2-12 were synthesized in
this laboratory as previously reported.24 Reaction of
chloride25 15 with aniline gave 1-oxide 1626 which was
oxidized with MCPBA to give 1,4-dioxide 13 (Scheme
1). Similarly, reaction of chloride 15 with aniline 18,
prepared from nitrobenzene 17,27 gave the 1-oxide 19
that was oxidized with trifluoroperacetic acid to the 1,4-
dioxide 14.
Figure 1. HT29 cells grown on a collagen-coated Teflon
microporous support membrane (Biopore, average thickness
Methods
30 µm) in Millicell-CM cell culture inserts (Millipore) (H & E
stained frozen section). Drug diffusion is measured in a
diffusion chamber that separates two compartments (donor
and receiver) that each hold about 5.5 mL of well-stirred
culture medium at 37 °C.
Partition Coefficients. The octanol-water partition
coefficients (P) of TPZ (1) and 2-14 were measured
using the shake-flask method,28 with analysis by HPLC.
MCL Cultures. MCLs were grown from human
HT29 colon carcinoma cells as described elsewhere.18,29
In brief, 1 × 106 cells were seeded on collagen-coated
Teflon microporous support membranes in Millicell-CM
cell culture inserts and grown for 3 days submerged in
stirred culture medium supplemented with 10% fetal
calf serum.
Determination of Diffusion Coefficient in MCLs.
Flux through MCLs was measured in a diffusion
chamber as described.23 TPZ analogues were added to
the “donor” compartment to 50 µM, along with the flux
markers [14C]urea, D-[2-3H]mannitol, and 9(10H)-acri-
done. The diffusion of each compound was also deter-
mined in the collagen-coated Teflon porous support
membrane without an MCL present. Samples were
taken from both the donor and receiver compartment
for up to 5 h. An aliquot was assayed for radioactivity
by dual-label liquid scintillation counting, and the
balance was analyzed by HPLC. The concentration-
time profiles were numerically fitted to Fick’s second
law to derive DMCL with addition of reaction terms in
the MCL when necessary, and the effective diffusion
coefficient in the support membrane (Ds).23,23 The latter
parameter reflects diffusion through the aqueous pores
in the support membrane but also takes into account
the unstirred boundary layers on each side of the
membrane.
tissue-like cell densities. These 3D structures eventually
become diffusion-limited, with central hypoxia and
necrosis,15,21 and thus mimic the extravascular compart-
ment in tumors. The simple planar geometry of MCLs
allows direct measurement of drug transport by quan-
tifying the flux of drug across an MCL that separates
two well-stirred compartments in a diffusion chamber.23
The concentrations of the drug and its metabolites, in
the “donor” and “receiver” compartments, can be fitted
to Fick’s second law (with added reaction terms for
reversible binding and/or metabolism when appropriate)
to provide the apparent diffusion coefficient in the MCL
(DMCL). This parameter can then be used, in conjunction
with the plasma pharmacokinetics (PK), to develop a
distributed parameter (i.e., spatially resolved) PK model
that describes the concentration-time profile of the
drug at each position within the extravascular compart-
ment. This PK model is coupled to a pharmacodynamic
(PD) model that describes the biological response (e.g.,
tumor cell kill) as a function of concentration and time,
allowing prediction of activity against cells as a function
of distance from blood vessels.18,23
In this study, we examine the influence of lipophilicity
on the diffusion of a series of neutral analogues of TPZ
in MCL cultures grown from the human colon carcinoma
cell line HT29. To minimize the effects of bioreductive
drug metabolism, the experiments were conducted using
high ambient oxygen concentrations (95% in the gas
phase), which reoxygenates the central hypoxic zone in
MCLs.15 We use this approach to define the relationship
between log P and DMCL. In addition, we use a distrib-
Results and Discussion
The relationship between octanol-water partition
coefficient P and the diffusion coefficient in MCLs (DMCL
)
was investigated for TPZ and the 13 analogues shown

Extravascular Transport of Tirapazamine Analogues
Journal of Medicinal Chemistry, 2005, Vol. 48, No. 4 1081
Table 1. Structures of Compounds, Octanol-Water Partition Coefficients (P), and Diffusion Coefficients in Teflon Support
Membranes (Ds) and HT29 MCLsa (DMCL
)
b
b
compd
formula
R
Mr (Da)
log Pb
Ds (×10-6 cm2 s-1
)
DMCL (×10-6 cm2 s-1
)
acridone
195
182
60
2.975 ( 0.006 (52)
-2.616 ( 0.002 (6)
-2.01 ( 0.01 (6)
-0.342 ( 0.005 (6)
0.068 ( 0.007 (3)
0.127 ( 0.001 (3)
0.24 ( 0.01 (3)
1.12 ( 0.02 (34)
1.22 ( 0.03 (6)
3.17 ( 0.05 (60)
0.21 ( 0.01 (32)
0.45 ( 0.02 (6)c
0.40 ( 0.01 (12)c
0.54 ( 0.03 (3)
0.66 ( 0.02 (2)
0.75 ( 0.02 (2)
0.21 ( 0.03 (2)
0.32 ( 0.03 (2)
0.35 ( 0.01 (2)
0.87 ( 0.02 (2)
0.67 ( 0.04 (2)
0.90 ( 0.06 (2)
0.52 ( 0.01 (2)
0.45 ( 0.02 (2)
2.70 ( 0.06 (2)
2.40 ( 0.01 (2)
mannitol
urea
2.08 ( 0.05 (19)c
1.26 ( 0.04 (5)c
1.38
1 (TPZ)
I
178
192
192
192
256
256
212
212
212
246
246
249
254
312
2
I
6-Me
3
I
7-Me
1.2 ( 0.1 (2)
4
I
8-Me
1.32 ( 0.06 (2)
1.22 ( 0.08 (2)
1.15 ( 0.02 (2)
1.05 ( 0.09 (2)
1.22 ( 0.08 (2)
1.3 ( 0.1 (2)
5
I
7-SO2Me
8-SO2Me
5-Cl
-1.15 ( 0.02 (3)
-1.382 ( 0.006 (3)
-0.102 ( 0.008 (3)
0.441 ( 0.003 (3)
0.166 ( 0.002 (3)
0.82 ( 0.01 (3)
6
I
7
I
8
I
7-Cl
8-Cl
7-CF3
8-CF3
8-NEt2
9
I
10
11
12
13
14
I
1.19 ( 0.01 (2)
1.30 ( 0.09 (2)
0.86 ( 0.02 (2)
1.01 ( 0.09 (2)
0.83 ( 0.02 (2)
I
0.48 ( 0.01 (3)
I
0.830 ( 0.003 (3)
1.425 ( 0.005 (6)
1.508 ( 0.007 (3)
II
III
a MCL ) multicellular layer. b Mean ( SEM or range with number of measurements in parentheses. c From Hicks et al.18
Figure 2. Mr dependence of diffusion coefficients (Ds) in
collagen-coated Teflon membranes used to support MCL
growth. Flux was measured in RMEM culture medium supple-
mented with 10% fetal calf serum at 37 °C. The compounds
(including the flux markers) of Table 1 are denoted by the open
symbols. The line is a linear regression, with slope -0.37.
Figure 3. Concentrations of TPZ (1, closed symbols) and
compound 14 (open symbols) as a function of time in the donor
(circles and down triangles) and receiver (squares and up
triangles) compartments during a representative flux experi-
ment with HT29 MCLs (solid lines) or Teflon support mem-
branes (dashed lines) under 95% O2 at 37 °C. Concentrations
(C) were normalized against the initial concentration in the
donor compartment (C0; 50 µM). The lines are fits to a simple
Fickian diffusion model, based on the average thickness of each
MCL as estimated from the flux of the internal standard urea
(not shown). Flux of 14 was faster than for TPZ, despite a
slightly greater thickness of the MCL used for 14 (167 µm)
than for TPZ (149 µm).
in Table 1, along with three internal standards that
were included in all experiments. The standards were
mannitol (a marker of the paracellular transport route),
9(10H)-acridone (a marker of the transcellular route),
and urea (used to determine thickness of each MCL).
The measured values of P for these 17 compounds
spanned almost 6 orders of magnitude.
To interpret the MCL flux data, we first determined
the effective diffusion coefficient of each compound
through the collagen-coated Teflon support membranes
(Ds) on which the MCLs are grown (Figure 2). These
membranes have previously been shown to have an
effective porosity of ca. 11%;23 this estimate includes the
effect of unstirred boundary layers. Given their large
pore size (0.4 µm), Ds is therefore expected to be
approximately 11% of the diffusion coefficient in culture
medium for all compounds. In each case, the flux curves
(concentration-time profiles in the two compartments
of the diffusion chamber as illustrated for TPZ and
compound 14 in Figure 3) were well-fitted by a simple
Fickian diffusion model, with no significant loss of total
drug over at least 5 h, confirming chemical stability and
lack of binding to any components of the apparatus.
Fitted values of Ds showed little variation between
compounds (Table 1), as expected, given the narrow
range of Mr. Taking a broader set of compounds (includ-
3
ing H2O, the internal standards, and 16 other com-
pounds chemically unrelated to TPZ such as quaternary
ammonium salts, dinitrobenzamides, and DNA inter-
calators), a significant relationship could be demon-
strated between Ds and Mr (Figure 2). The regression
line in Figure 2 (N ) 34, R2 ) 0.81, F ) 136) has a slope
of -0.37, indicating that Ds is inversely proportional to
Mr0.37. This is not significantly different from the

1082 Journal of Medicinal Chemistry, 2005, Vol. 48, No. 4
Pruijn et al.
theoretical dependence (Mr0.33) for spherical molecules
in water according to the Stokes-Einstein equation.30
In contrast to the narrow range of Ds values for the
TPZ analogues, diffusion through MCLs differed con-
siderably between compounds. This is illustrated for
TPZ and a lipophilic analogue (13) in Figure 3; although
the latter compound diffused slightly more slowly than
TPZ through bare support membranes (Figure 2), it
diffused much more rapidly through MCLs. For both
compounds, mass balance was preserved (the total
amount of compound in both compartments did not
change with time), no metabolites were detected in the
chromatograms, and the flux curves were well fitted as
simple diffusion. Thus, metabolism in the MCLs was
not a factor for these compounds at the high O2
concentrations used in this study. It is important to note
that TPZ is not a known substrate for drug transporters
such as P-gp or MRP1 or -2 and that HT29 cells do not
express significant levels of these proteins;31 thus simple
passive diffusion is the expected transport mode.
This same approach was used to investigate the other
TPZ analogues. Most were also metabolically stable in
HT29 MCLs, but six of the compounds with electron-
withdrawing substituents (7- and 8-SO2Me, 5- and 8-Cl,
7- and 8-CF3) were metabolized significantly (negative
mass balance and metabolite peaks in the chromato-
grams; in some cases these metabolites were identified
as the corresponding 1-oxides resulting from two-
electron reduction of the 4-oxide moiety by DT-diapho-
rase; F. B. Pruijn and A. V. Patterson, unpublished
data). For these compounds, diffusion coefficients could
be estimated by fitting a first-order metabolism term
and DMCL simultaneously. The rate constants for me-
tabolism within the MCLs were 0.24 min-1 for 5, 0.44
min-1 for 6, 0.07 min-1 for 7, 3.41 min-1 for 9, 0.35
min-1 for 10, and 1.84 min-1 for 11.
Figure 4. The log P dependence of diffusion coefficients in
HT29 MCLs, corrected for differences in molecular size (D′MCL
,
corrected to the same Mr as TPZ (1)) for flux markers (open
symbols) and TPZ analogues (closed symbols). Error bars (see
Table 1) have been omitted for clarity; numbers refer to TPZ
analogues as in Table 1. The line is the fit to a modified form
of eq 1 in which the Mr dependence is removed by normalizing
to the Mr of TPZ (R ) 0.941, F ) 33.47, SE ) 0.138).
to that observed in aqueous media (Mr-0.37) and in good
agreement with the observed molecular weight depen-
dence of the permeability surface area product of animal
muscle capillary wall (Mr-0.51) and of human muscle
(Mr-0.50).32
To clarify the log P dependence of DMCL, we normal-
ized for differences in size of the molecules by calculat-
ing the expected diffusion coefficient at the same Mr as
TPZ (D′MCL), using the Mr dependence determined in
eq 1. These size-corrected values, plotted in Figure 4,
showed an increase from approximately 0.3 × 10-6 cm2
s-1 for hydrophilic molecules to a 15-fold higher value
for the most lipophilic examples. The latter is ap-
proximately 3-fold lower than D′ in water (calculated
using D of urea in water30 as 8.76 × 10-6 cm2 s-1 at an
Mr of 178 Da). The dependence on log P is interpreted
as reflecting a transition from paracellular to transcel-
lular transport, with the more hydrophilic analogues
effectively restricted to the extracellular diffusion path.
These findings are similar to those reported for passive
diffusion in epithelial monolayers (e.g., Caco-2). For
example, a sigmoidal curve was obtained for the loga-
rithm of the apparent permeability coefficient of â-block-
ers against the logarithm of the distribution coefficient
(i.e., lipophilicity) in Caco-2 monolayers,33,34 which was
interpreted in terms of the relative contributions of
trans- and paracellular diffusion barriers.34 Such an
experimental relationship can also be derived from
theoretical considerations.35 The highest achievable
DMCL (i.e., the asymptote of the sigmoidal function) is
presumably determined mainly by the microviscosity of
cytosol but also to some extent of lipoidal membranes
and extracellular matrix and is within the range of 2-
to 5-fold lower diffusion coefficient for small molecules
in cytoplasm in a variety of cell systems.36
The values of DMCL obtained by modeling the fluxes
as diffusion (with metabolism when appropriate) are
shown in Table 1. Regression analysis demonstrated a
statistically significant relationship between DMCL and
log P for all the compounds of Table 1, using a semiem-
pirical three-parameter sigmoidal function (eq 1; R )
0.933, F ) 20.25, SE ) 0.143),
log(DMCL) ) C - [a log(Mr)] + f(log P)
(1)
where
R
f(log P) )
1 + e-(log P-â)/γ
and where C is a constant and a, R, â, and γ are
regression coefficients. The fitted values were C ) -5.6,
a ) 0.5, R ) 1.2, â ) 0.7, and γ ) 0.6. Inclusion of the
Mr term reduced the variance slightly (with the R value
increasing from 0.917 when a ) 0), but the F-statistic
for addition of this term was only 2.80, and this was
not statistically significant (P ) 0.12). Although the
overall statistics of eq 1 did not improve significantly
on adding the Mr term (with this set of compounds
falling within a relatively narrow size range), there are
strong a priori reasons to include this term since on
theoretical grounds one would expect a size-dependence
at least as marked as for Ds (vide supra). Consistent
with expectation, the Mr dependence (Mr-0.5) was similar
The log P dependence in the present study indicates
that a large (10-fold) increase in the effective diffusion
coefficient relative to TPZ in tissue can be achieved by
designing compounds with sufficient lipophilicity to
permeate plasma membranes efficiently. We next asked
whether the observed changes in DMCL are large enough
to have a major effect on the ability of TPZ analogues

Extravascular Transport of Tirapazamine Analogues
Journal of Medicinal Chemistry, 2005, Vol. 48, No. 4 1083
increasing the rate of metabolic activation 5-fold. The
increased rate of metabolism actually reduces hypoxic
cell kill in the case of a compound with the same
diffusion coefficient as TPZ, due to decreased penetra-
tion, but improves hypoxic cell killing by almost 10 logs
for the fast-diffusing analogue. We have previously
noted that there will be an optimum rate of metabolism
for TPZ analogues since a rate that is too rapid impedes
penetration and a rate that is too slow limits production
of the cytotoxic radical reduction product.15 The PK/PD
model indicates that the optimum rate of metabolism
would be shifted to higher values if the diffusion
coefficient were increased and that attempts to develop
improved analogues by increasing rates of metabolic
reduction alone are likely to fail. The rate of metabolism
can be modulated by electron-withdrawing or -donating
substituents in the A-ring, for example, as noted above
with 5, 6, 7, 9, 10, and 11, and this will also have an
impact on hypoxia-selective cytotoxic potency.24,38,39 We
conclude that increasing tissue diffusion coefficients by
increasing drug lipophilicity has the potential to provide
considerably improved TPZ analogues by diminishing
the constraint imposed by limited extravascular trans-
port; this would allow development of compounds that
are more rapidly activated by enzymatic reduction
under hypoxia and would provide much greater activity
against hypoxic cells in tumors.
Figure 5. Sensitivity of hypoxic cell killing in tumors to
changes in diffusion coefficient of TPZ analogues, simulated
using a spatially resolved PK/PD model as described in the
Experimental Section. The solid lines are based on the
measured rate of metabolism of TPZ (1) by HT29 cells and
the dashed line on a 5-fold higher rate of drug metabolism.
(A) Steady-state concentration gradients as fractions of the
plasma concentration (Cp) for O2 (Cp ) 50 µM), TPZ, and a TPZ
analogue with 6-fold higher diffusion coefficient (as for 14)
assuming Cp ) 40 µM and the same rate of metabolism for
both compounds. (B) Predicted cell killing for the same two
compounds. The PD parameters for the faster-diffusing ana-
logue are assumed to be the same as for TPZ.
to kill hypoxic cells in tumors. To address this question,
we used a distributed parameter PK/PD model for TPZ
to simulate the effect of changing DMCL; the formalism
of this model is outlined in the Experimental Section.
Briefly, the PK/PD model used the MCL transport
parameters to predict PK as a function of distance from
blood vessels and used PD parameters (relationship
between drug exposure, O2 concentration, and killing)
determined using HT29 single cell suspensions.18,37 This
model follows that reported earlier15 except that rather
than representing the tumor as comprising an oxic and
anoxic compartment it models the O2 gradient from
blood vessels and incorporates the O2 dependence of
metabolism and cytotoxicity.37 Key aspects of this PK/
PD model have been validated recently by showing that
it correctly predicts the apparent resistance to TPZ of
anoxic HT29 cells in MCLs relative to single cells in
suspension.18
The predictions of the PK/PD model are shown in
Figure 5. The steady-state concentration gradient of
TPZ is not as steep as for O2, but the TPZ concentration
in the most hypoxic regions is lowered to 62% of that in
the perivascular region because of its metabolic con-
sumption during diffusion (Figure 5A). The PK/PD
model demonstrates the expected increase in killing
with distance (Figure 5B) because of selective activation
under hypoxia. When the DMCL is increased 6-fold from
0.4 × 10-6 to 2.4 × 10-6 cm2 s-1 (i.e., to the value for
14) without changing the rate of TPZ metabolism, the
steady-state concentration gradient would be flattened
further (analogue in Figure 5A); as a consequence,
hypoxic cell killing would be expected to increase, with
the surviving fraction of the most hypoxic cells falling
from 0.0017 to 2.7 × 10-6 (Figure 5B). Thus, increasing
the lipophilicity of TPZ analogues provides a strategy
for minimizing the extravascular penetration problem
and thereby improving activity against target cells. The
gains from increasing DMCL are even greater if the rate
of bioreductive metabolism (and hence drug potency) is
also increased. Thus, also shown in Figure 5 (as dashed
lines) is the effect on exposure and cytotoxicity of
Whether the relationship between log P and DMCL for
TPZ analogues can be generalized to other classes of
antitumor agents remains to be determined. Initial
investigation of analogues with 3-alkyl substituents
(replacing the 3-NH2 group of TPZ) suggests that these
may fall on a different line, with higher values of DMCL
,
possibly because the 3-NH2 group of TPZ impedes
membrane permeation as a consequence of its H-bond
donor/acceptor properties. It has been reported for
peptide and peptide mimetics that passive membrane
permeability is best described by a combination of terms
for lipophilicity and hydrogen bond potential.40,41 How-
ever, even if the relationship observed for TPZ ana-
logues is not universal, the present study clearly
demonstrates the importance of drug lipophilicity as a
determinant of extravascular transport and provides a
general methodology for investigating this relationship
within congeneric drug series.
Conclusions
The flux of 17 neutral compounds (primarily ana-
logues of the hypoxic cytotoxin tirapazamine) across
multicellular layers of HT29 human tumor cells has
been determined, and the diffusion coefficients (DMCL
)
were derived. We demonstrate a marked sigmoidal
dependence of DMCL on drug lipophilicity and show that
molecular weight is of lesser importance over the size
range of typical small molecule drugs. There is some
evidence from preliminary data that hydrogen bonds
may also affect DMCL. The demonstrated structure-
activity relationships provide a strategy for improving
the extravascular transport of tirapazamine through the
development of more lipophilic analogues. PK/PD mod-
eling shows that the increase in DMCL achievable by
raising log P is sufficient to give a large increase in
cytotoxicity against hypoxic cells in tumors. This gain
is even greater if the kinetics of metabolic activation of

1084 Journal of Medicinal Chemistry, 2005, Vol. 48, No. 4
Pruijn et al.
was chromatographed, eluting with 10% EtOAc/pet. ether, to
give 1-oxide 16 (334 mg, 49%) as a yellow powder: mp 197-
198.5 °C (lit.26 199-201 °C); 1H NMR δ 8.32 (d, J ) 9.0 Hz, 1
H, H-8), 7.70-7.77 (m, 4 H, H-5, H-6, H-2′, H-6′), 7.37-7.42
(m, 3 H, H-7, H-3′, H-5′), 7.22 (br s, 1 H, NH), 7.13 (dt, J )
7.5, 0.9 Hz, 1 H, H-4′); 13C NMR δ 156.3, 148.1, 138.1, 135.8,
131.6, 129.1 (2), 127.1, 126.1, 123.8, 120.4, 119.7 (2).
the analogue is increased to take full advantage of the
increased rate of diffusion. Assuming that the extravas-
cular transport problem is more severe in tumors than
in normal tissue (as expected, given the superior mi-
crovascular networks and relative lack of hypoxia in the
latter),42 this is expected to provide analogues with
improved therapeutic ratios in vivo.
N-Phenyl-1,2,4-benzotriazin-3-amine 1,4-Dioxide (13).
A solution of MCPBA (0.99 g, 4.0 mmol) in DCM (10 mL) was
added to a stirred solution of 1-oxide 16 (0.64 g, 2.7 mmol) in
DCM (100 mL), and the solution was stirred at 20 °C for 2 h.
The solvent was evaporated, and the residue was chromato-
graphed, eluting with a gradient (10-20%) of EtOAc/DCM, to
give (i) starting material (0.34 g, 50%) spectroscopically
identical with the sample prepared above and (ii) 1,4-dioxide
Inefficient extravascular transport has been identified
as one of the key problems in cancer chemotherapy,43,44
but a lack of experimental models has made it difficult
to address this requirement explicitly. The MCL model
is potentially a useful tool for determining extravascular
transport properties of hits from drug discovery pro-
grams and for evaluating this requirement during lead
optimization. If the structure-activity relationship
demonstrated here for tirapazamine analogues can be
generalized to other structural types (and other cell
lines), then in silico prediction of extravascular trans-
port properties should be achievable. This would provide
a powerful tool for assisting with a critical (and ne-
glected) aspect of lead optimization, although the struc-
tural requirements for efficient extravascular transport
will need to be expressed within the additional con-
straints imposed by target interaction (prodrug activa-
tion/receptor binding), systemic pharmacokinetics, and
toxicology.
1
13 (85 mg, 12%) as a red powder: mp 201-203 °C; H NMR
δ 9.17 (br s, 1 H, NH), 8.37-8.41 (m, 2 H, H-5, H-8), 7.90-
7.94 (m, 1 H, H-6), 7.58-7.68 (m, 3 H, H-7, H-3′, H-5′), 7.43
(dd, J ) 7.8, 7.4 Hz, 2 H, H-2′, H-6′), 7.21 (t, J ) 7.4 Hz, 1 H,
H-4′); 13C NMR δ 147.5, 138.0, 136.1, 135.7, 131.3, 129.4 (2),
127.9, 125.2, 121.9, 120.8 (2), 117.8. Anal. (C13H10N4O2) C, H,
N.
N-[3-(2-Methoxyethyl)phenyl]-1,2,4-benzotriazin-3-
amine 1,4-Dioxide (14).
1-(2-Methoxyethyl)-3-nitrobenzene (17). A solution of
3-nitrophenethyl alcohol (1.05 g, 6.3 mmol) in THF (10 mL)
was added dropwise to a stirred suspension of NaH (325 mg,
8.1 mmol) in THF (30 mL) at 5 °C, and the mixture was
warmed to 20 °C and stirred 30 min. Iodomethane (3.9 mL,
62.5 mmol) was added, and the mixture was stirred at 20 °C
for 16 h. The solvent was evaporated and the residue parti-
tioned between EtOAc (100 mL) and water (100 mL). The
organic fraction was washed with water (2 × 30 mL) and brine
(30 mL) and dried, and the solvent was evaporated. The
residue was chromatographed, eluting with 20% EtOAc/pet.
ether, to give ether 17 (981 mg, 87%) as a clear oil:27 1H NMR
δ 8.06-8.11 (m, 2 H, H-2, H-4), 7.57 (d, J ) 7.6 Hz, 1 H, H-6),
7.47 (dd, J ) 7.9, 7.6 Hz, 1 H, H-5), 3.65 (t, J ) 6.5 Hz, 2 H,
Experimental Section
Chemistry. Analyses were carried out in the Microchemical
Laboratory, University of Otago, Dunedin, New Zealand.
Melting points were determined on an Electrothermal 2300
melting point apparatus. NMR spectra were obtained on a
Bruker AM-400 spectrometer at 400 MHz for 1H and 100 MHz
for 13C spectra. Spectra were obtained in CDCl3 unless
otherwise specified and are referenced to Me4Si. Chemical
shifts and coupling constants were recorded in units of ppm
and Hz, respectively. Assignments were determined using
COSY, HSQC, and HMBC two-dimensional experiments. Mass
spectra were determined on a VG-70SE mass spectrometer
using an ionizing potential of 70 eV at a nominal resolution of
1000. High-resolution spectra were obtained at nominal
resolutions of 3000, 5000, or 10 000 as appropriate. All spectra
were obtained as electron impact (EI) using PFK as the
reference unless otherwise stated. Solutions in organic solvents
were dried with anhydrous Na2SO4, and solvents were evapo-
rated under reduced pressure on a rotary evaporator. Thin-
layer chromatography was carried out on aluminum-backed
silica gel plates (Merck 60 F254) with visualization of compo-
nents by UV light (254 nm) or exposure to I2. Column
chromatography was carried out on silica gel (Merck 230-400
mesh). All compounds designated for testing were analyzed
at >99% purity by reverse phase HPLC using an Agilent 1100
liquid chromatograph, an Alltima C18 (5 µm) stainless steel
column (150 mm × 3.2 mm i.d.), and an Agilent 1100 diode
array detector. Chromatograms were run using various gra-
dients of aqueous (0.045 M ammonium formate and formic acid
at pH 3.5) and organic (80% MeCN/MilliQ water) phases. DCM
refers to dichloromethane; ether refers to diethyl ether; DME
refers to dimethoxyethane; EtOAc refers to ethyl acetate;
EtOH refers to ethanol; MeOH refers to methanol; pet. ether
refers to petroleum ether, boiling range 40-60 °C; THF refers
to tetrahydrofuran dried over sodium benzophenone ketyl. All
solvents were freshly distilled. Compounds were diluted 100-
fold from stock solutions (5 mM) in DMSO prior to use.
N-Phenyl-1,2,4-benzotriazin-3-amine 1,4-Dioxide (13).
N-Phenyl-1,2,4-benzotriazin-3-amine 1-Oxide (16). Two
drops of concentrated HCl were added to a solution of chloride
1525 (0.52 g, 2.86 mmol) and aniline (0.78 mL, 8.59 mmol) in
DME (10 mL), and the solution was stirred at reflux temper-
ature for 16 h. The solvent was evaporated, and the residue
CH2O), 3.36 (s, 3 H, OCH3), 2.98 (t, J ) 6.5 Hz, 2 H, CH2); 13
C
NMR δ 148.3, 141.3, 135.2, 129.2, 123.7, 121.4, 72.5, 58.8, 35.8.
3-(2-Methoxyethyl)aniline (18). A solution of ether 17
(928 mg, 5.1 mmol) in EtOH (50 mL) with Pd/C (100 mg) was
stirred under H2 (60 psi) for 2 h. The mixture was filtered
through Celite and washed with EtOH (2 × 10 mL), and the
solvent was evaporated to give aniline 18 (718 mg, 93%) as a
pale pink oil: 1H NMR δ 7.08 (dd, J ) 7.7, 7.3 Hz, 1 H, H-5),
6.62 (br d, J ) 7.3 Hz, 1 H, H-4), 6.51-6.55 (m, 2 H, H-2, H-6),
3.50 (br s, 2 H, NH2), 3.58 (t, J ) 7.2 Hz, 2 H, CH2O), 3.35 (s,
3 H, OCH3), 2.80 (t, J ) 7.2 Hz, 2 H, CH2); 13C NMR δ 146.4,
140.1, 129.3, 119.1, 115.7, 113.1, 73.6, 58.6, 36.2; MS (EI+) m/z
151 (M+, 90%), 136 (20), 106 (100); HRMS calcd for C9H13NO
(M+) m/z 151.0997; found, 151.0995.
N-[3-(2-Methoxyethyl)phenyl]-1,2,4-benzotriazin-3-
amine 1-Oxide (19). A solution of chloride 15 (376 mg, 2.1
mmol) and amine 18 (688 mg, 4.6 mmol) in DMSO (20 mL)
was heated at 100 °C for 16 h. The solution was partitioned
between EtOAc (100 mL) and water (100 mL), the organic
fraction was washed with water (2 × 50 mL) and brine (50
mL) and dried, and the solvent was evaporated. The residue
was chromatographed, eluting with a gradient (20-50%) of
EtOAc/pet. ether, to give 1-oxide 19 (590 mg, 96%) as an
orange powder: mp (EtOAc/Et2O) 122-124 °C; 1H NMR
[(CD3)2SO] δ 10.18 (s, 1 H, NH), 8.22 (dd, J ) 8.6, 1.0 Hz, 1 H,
H-8), 7.87 (ddd, J ) 8.5, 7.1, 1.3 Hz, 1 H, H-6), 7.70-7.76 (m,
3 H, H-5, H-2′, H-6′), 7.47 (ddd, J ) 8.6, 7.1, 1.3 Hz, 1 H, H-7),
7.27 (dd, J ) 7.9, 7.8 Hz, 1 H, H-5′), 6.94 (d, J ) 7.8 Hz, 1 H,
H-4′), 3.58 (t, J ) 6.8 Hz, 2 H, CH2O), 3.27 (s, 3 H, OCH3),
2.82 (t, J ) 6.8 Hz, 2 H, CH2); 13C NMR [(CD3)2SO] δ 156.3,
147.5, 139.5 (C-1′), 139.1 (C-4a), 135.9 (C-6), 130.9 (C-8a), 128.4
(C-5′), 126.6 (C-5), 125.8 (C-4′), 123.2, 119.8, 119.7, 117.3, 72.6,
57.8, 35.5. Anal. (C16H16N4O2) C, H, N.
N-[3-(2-Methoxyethyl)phenyl]-1,2,4-benzotriazin-3-
amine 1,4-Dioxide (14). H2O2 (70%, 260 µL, ca. 5.2 mmol)
was added dropwise to a stirred solution of trifluoroacetic

Extravascular Transport of Tirapazamine Analogues
Journal of Medicinal Chemistry, 2005, Vol. 48, No. 4 1085
anhydride (730 µL, 5.2 mmol) in DCM (5 mL) at 5 °C. The
solution was stirred at 5 °C for 10 min, then warmed to 20 °C
and added dropwise to a stirred solution of 1-oxide 19 (307
mg, 1.0 mmol) in CHCl3 (10 mL) at 20 °C. The red solution
was stirred for 5 h, diluted with water (10 mL), and extracted
with CHCl3 (4 × 10 mL), the combined organic fraction was
dried, and the solvent was evaporated. The residue was
chromatographed, eluting with a gradient of (0-5%) of MeOH/
(50-100%) EtOAc/pet. ether, to give (i) starting material 19
(143 mg, 47%) spectroscopically identical with the sample
prepared above and (ii) 1,4-dioxide 14 (62 mg, 19%) as a red
solid: mp (EtOAc/Et2O) 148-149 °C; 1H NMR [(CD3)2SO] δ
10.11 (s, 1 H, NH), 8.23-8.28 (m, 2 H, H-5, H-8), 8.00 (ddd, J
) 8.5, 7.2, 1.0 Hz, 1 H, H-6), 7.65 (ddd, J ) 8.6, 7.2, 0.8 Hz, 1
H, H-7), 7.50-7.53 (m, 2 H, H-2′, H-6′), 7.31 (t, J ) 8.7, 7.6, 1
H, H-5′), 7.05 (d, J ) 7.6 Hz, 1 H, H-4′), 3.57 (t, J ) 6.8 Hz, 2
H, CH2O), 3.26 (s, 3 H, OCH3), 2.82 (t, J ) 6.8 Hz, 2 H, CH2);
13C NMR [(CD3)2SO] δ 147.7, 139.6, 138.2, 138.8, 135.5, 131.0,
128.4, 127.7, 124.7, 122.4, 121.1, 118.7, 117.3, 72.5, 57.7, 35.3.
Anal. (C16H16N4O3) C, H, N.
MCL Cultures. MCLs were grown from human HT29 colon
carcinoma cells as described elsewhere.18,29 In brief, 1 × 106
cells were seeded on collagen-coated Teflon microporous sup-
port membranes (Biopore; average thickness 30 µm) in Milli-
cell-CM cell culture inserts (Millipore Corporation, Bedford,
MA) and grown for 3 days submerged in stirred culture
medium (RMEM; GIBCO BRL, Grand Island, NY) supple-
mented with 10% heat-inactivated fetal calf serum (GIBCO
BRL, Auckland), penicillin (100 U/mL), streptomycin (100 µg/
mL), and 2 mM l-glutamine.
Determination of Diffusion Coefficient in MCLs. Flux
through MCLs was measured in a diffusion chamber as
described,23 using MCLs equilibrated for 60 min with 95% O2/
5% CO2 at 37 °C in the same medium as above. TPZ analogues
were added to the “donor” compartment to 50 µM, along with
[14C]urea (Amersham Pharmacia Biotech; 40 MBq/mmol, 7.5
kBq/mL), D-[2-3H]mannitol (ICN Pharmaceuticals Inc., Irvine,
CA; 40 MBq/mmol, 20 kBq/mL), and 9(10H)-acridone (Sigma-
Aldrich, Castle Hill, NSW; 10 µM). Samples were taken from
both the donor and receiver compartment for up to 5 h. An
aliquot was assayed for radioactivity by dual-label liquid
scintillation counting in a Packard Tri-Carb 1500 liquid
scintillation analyzer (Packard Instrument Company, Meriden,
CT) using Emulsifier-Safe water-accepting scintillant (Pack-
ard), and the balance was frozen for subsequent HPLC
analysis. The concentration-time profiles of [14C]urea in the
donor and receiver compartments were numerically fitted to
Fick’s second law to estimate the average thickness of each
MCL (ca. 175 µm), using the measured value of DMCL for [14C]-
urea in HT29 MCLs (0.45 × 10-6 cm2 s-1).18 The effective
diffusion coefficient of each compound was also determined in
the collagen-coated Teflon porous support membrane (Ds)
without an MCL present. This latter parameter also takes into
account the effect of the unstirred boundary layers on each
side of the membrane or MCL. DMCL was then determined by
fitting the concentration-time profile in both the donor and
receiver simultaneously to a Fickian diffusion model with the
support membrane and MCL in series as described previ-
ously,23 with addition of reaction terms in the MCL when
necessary.
HPLC. Samples (200 µL) were deproteinized by addition
of 4 µL of 70% (v/v) perchloric acid and chilling on ice followed
by centrifugation (12000×g for 5 min at 4 °C) and subsequent
neutralization of the supernatant with 50% (v/v) ammonia
(31.5 µL/mL supernatant). Concentrations of the TPZ ana-
logues were determined by reverse phase HPLC (Alltima C8 5
µm column, 150 mm × 2.1 mm; Alltech Associated Inc.,
Deerfield, IL) using an Agilent HP1100 equipped with a diode-
array detector. Mobile phases were gradients of 80% acetoni-
trile/20% H2O (v/v) in 450 mM ammonium formate, pH 4.5,
at 0.3 mL/min. Quantitation was based on calibration curves
in mobile phase (0.1-100 µM), corrected for recovery from
medium with serum determined by assaying known concen-
trations (0.1-100 µM) for each compound under the same
conditions.
Partition Coefficients. The octanol-water partition coef-
ficient (P) was measured using the shake-flask method,28 using
GPR-grade octanol (BDH Laboratory Supplies). Briefly, lipo-
philic drugs were dissolved directly in octanol-saturated PBS
(137 mM NaCl, 2.68 mM KCl, 1.47 mM KH2PO4, 8.10 mM Na2-
HPO4, pH 7.4), and hydrophilic drugs were dissolved in PBS-
saturated octanol to 25-100 µM. Equal volumes of PBS and
octanol were mixed on a Bellco roller drum (Bellco Glass, Inc.,
NJ) at 20 rpm for 3 h at ambient temperature. The two solvent
layers were separated after a brief spin and analyzed by HPLC
directly (aqueous layer) or after addition of 4 volumes of
methanol (organic layer).
Mathematical Modeling. Using the transport parameters
governing extravascular diffusion derived from MCL flux
studies and the plasma PK, we calculated the concentration-
time profile for TPZ and analogues as a function of position,
x, and time, t, in the extravascular compartment as previously
described.37 First, the oxygen diffusion gradient (which deter-
mines the rate of hypoxic drug metabolism) was calculated in
the same way as for TPZ (below) using representative litera-
ture values (diffusion coefficient 1.25 × 10-5 cm2 s-1 45
,
con-
sumption rate 0.022 cm3 g-1 min-1 46
input capillary concen-
,
tration of 50 µM [ca. 40 mmHg] O247). A planar geometry (i.e.,
bidirectional diffusion into a slab of tissue) was assumed as a
compromise between radial outward diffusion in a Krogh
cylinder and radial inward diffusion for a tumor cord sur-
rounded by stroma. Under these conditions the O2 concentra-
tion falls to 10 ppm at 95 µm from the plasma compartment.
It was also assumed that a small population of cells survives
transiently under this very low oxygen tension, up to 5 µm
beyond the diffusion limit of O2. This simulation gives a
radiobiologically hypoxic fraction (proportion of cells at <4 µM
O2) of 36%.
The recently determined kinetics of the metabolic activation
of TPZ18 and its oxygen dependence37 were used to simulate
TPZ metabolism in tissue as follows:
Vmax[TPZ]
∂M
∂t
K
)
kmet[TPZ] +
(2)
(
[O ] + K)(
)
Km + [TPZ]
2
where M(x,t) is the cumulative amount of TPZ metabolized
per unit intracellular volume, K is the oxygen concentration
required to halve the rate of TPZ metabolism (1.21 µM),37 and
kmet (0.78 min-1), Vmax (8.5 µM min-1), and Km (3.5 µM) are
the first-order and Michaelis-Menten parameters governing
TPZ metabolism in anoxic HT29 cells.18 The following reac-
tion-diffusion equation
∂[TPZ]
∂t
∂2[TPZ]
∂x2
∂M
∂t
) D
- φ
(3)
was then solved for [TPZ] at a distance x and time t using the
appropriate tissue diffusion coefficient (D), which is assumed
to be identical to the experimentally determined DMCL and
intracellular volume fraction (φ). The boundary conditions were
defined by the plasma PK profile in humans in recent clinical
trials48-50 where TPZ is infused over 2 h at a dose of 390 mg/
m2, as calculated using PK parameters from the phase I
studies.51,52 This would produce a maximum area under the
plasma concentration time curve (AUCp) of 150 µM h with a
plasma concentration of 40 µM over the first 2.5 h followed by
a first-order decay with a half-life of 47 min. This simulation
provided the drug PK profile as a function of distance within
the extravascular compartment, which rapidly reaches the
steady state profile shown in Figure 5A for the first 2.5 h and
then decays with a similar half-life to that of drug in plasma.
This was used in conjunction with a recently determined PD
model for TPZ cytotoxicity18 to calculate cell killing as a
function of distance. Briefly, the PD model assumes that the
log cell kill under steady-state conditions is proportional to

1086 Journal of Medicinal Chemistry, 2005, Vol. 48, No. 4
Pruijn et al.
(19) Kubinyi, H. Lipophilicity and biological activity. Arzneim.-
Forsch. 1979, 29, 1067-1080.
(20) Lieb, W. R.; Stein, W. D. Simple diffusion across the membrane
bilayer. In Transport and Diffusion across Cell Membranes;
Stein, W. D., Ed.; Academic Press: Orlando,FL, 1986; pp 69-
112.
the product of the TPZ concentration and the cumulative
metabolism of TPZ to its active radical:
-log(SF) ) γ×[TPZ]×M
(4)
(21) Cowan, D. S.; Hicks, K. O.; Wilson, W. R. Multicellular mem-
branes as an in vitro model for extravascular diffusion in
tumours. Br. J. Cancer, Suppl. 1996, 27, S28-S31.
(22) Minchinton, A. I.; Wendt, K. R.; Clow, K. A.; Fryer, K. H.
Multilayers of cells growing on a permeable support. An in vitro
tumour model. Acta Oncol. 1997, 36, 13-16.
(23) Hicks, K. O.; Pruijn, F. B.; Baguley, B. C.; Wilson, W. R.
Extravascular transport of the DNA intercalator and topo-
isomerase poison N-[2-(dimethylamino)ethyl]acridine-4-carboxa-
mide (DACA): diffusion and metabolism in multicellular layers
of tumor cells. J. Pharmacol. Exp. Ther. 2001, 297, 1088-1098.
(24) Hay, M. P.; Gamage, S. A.; Kovacs, M. S.; Pruijn, F. B.; Anderson,
R. F.; Patterson, A. V.; Wilson, W. R.; Brown, J. M.; Denny, W.
A. Structure-activity relationships of 1,2,4-benzotriazine 1,4-
dioxides as hypoxia-selective analogues of tirapazamine. J. Med.
Chem. 2003, 46, 169-182.
where SF is the surviving fraction and γ is the value (2.33 ×
10-5 µM-2) determined for HT29 cells.18 The cell kill predicted
by this model for the two compounds is shown in Figure 5B.
Acknowledgment. The authors thank Dr. Maruta
Boyd, Sisira Kumara, and Li Fong Leong for technical
support. This work was supported by Grant No. CA82566
from the U.S. National Cancer Institute (F.B.P., J.R.S.,
H.D.S.L., K.O.H., M.P.H.), the Health Research Council
of New Zealand (W.R.W.), and the Auckland Division
of the Cancer Society of New Zealand (W.A.D.).
Supporting Information Available: Elemental analysis
data for compounds 13, 14, 18, and 19. This material is
available free of charge via the Internet at http://pubs.acs.org.
(25) Robbins, R. F.; Schofield, K. Polyazabicyclic compounds. Part
II. Further derivatives of benzo-1,2,4-triazine. J. Chem. Soc.
1957, 3186-3194.
(26) Pazdera, P.; Potacek, M. N-Ethyl substituted 2-nitrophen-
ylguanidines. II Cyclization. Chem. Pap. 1989, 43, 107-112.
(27) Norman, R. O. C.; Radda, G. K. The ortho:para ratio in aromatic
substitution. Part 1. The nitration of methyl phenethyl ether.
J. Chem. Soc. 1961, 3030-3037.
(28) Siim, B. G.; Hicks, K. O.; Pullen, S. M.; van Zijl, P. L.; Denny,
W. A.; Wilson, W. R. Comparison of aromatic and tertiary amine
N-oxides of acridine DNA intercalators as bioreductive drugs.
Cytotoxicity, DNA binding, cellular uptake, and metabolism.
Biochem. Pharmacol. 2000, 60, 969-978.
(29) Wilson, W. R.; Pullen, S. M.; Hogg, A.; Hobbs, S. M.; Pruijn, F.
B.; Hicks, K. O. In vitro and in vivo models for evaluation of
GDEPT: quantifying bystander killing in cell cultures and
tumors. In Suicide Gene Therapy: Methods and Reviews;
Springer, C. J., Ed.; Humana Press: Totowa, NJ, 2003; pp 403-
432.
(30) Cussler, E. L. Values of diffusion coefficients. In Diffusion, Mass
Transfer in Fluid Systems, 2nd ed.; Cussler, E. L., Ed.; Cam-
bridge University Press: Cambridge and New York, 1997; pp
101-141.
References
(1) Fujimori, K.; Covell, D. G.; Fletcher, J. E.; Weinstein, J. N.
Modeling analysis of the global and microscopic distribution of
immunoglobulin G, F(ab′)2, and Fab in tumors. Cancer Res. 1989,
49, 5656-5663.
(2) van Osdol, W.; Fujimori, K.; Weinstein, J. N. An analysis of
monoclonal antibody distribution in microscopic tumor nod-
ules: consequences of a “binding site barrier”. Cancer Res. 1991,
51, 4776-4784.
(3) Rowe-Jones, D. C. The penetration of cytotoxins into malignant
tumours. Br. J. Cancer 1968, 22, 155-162.
(4) Jain, R. K.; Weissbrod, J. M.; Wei, J. Mass transport in tumors:
characterization and applications to chemotherapy. Adv. Cancer
Res. 1980, 33, 251-310.
(5) Blasberg, R.; Horowitz, M.; Strong, J.; Molnar, P.; Patlak, C.;
Owens, E.; Fenstermacher, J. Regional measurements of [14C]-
misonidazole distribution and blood flow in subcutaneous RT-9
experimental tumors. Cancer Res. 1985, 45, 1692-1701.
(6) Kerr, D. J.; Kaye, S. B. Aspects of cytotoxic drug penetration
with particular reference to anthracyclines. Cancer Chemother.
Pharmacol. 1987, 19, 1-5.
(7) Simpson-Herren, L.; Noker, P. E. Diversity of penetration of anti-
cancer agents into solid tumours. Cell Proliferation 1991, 24,
355-365.
(8) Jain, R. K. The next frontier of molecular medicine: delivery of
therapeutics. Nat. Med. 1998, 4, 655-657.
(9) Tannock, I. F. Conventional cancer therapy: promise broken or
promise delayed? Lancet 1998, 351 (Suppl. 2), 9-16.
(10) Brown, J. M. SR 4233 (tirapazamine): a new anticancer drug
exploiting hypoxia in solid tumours. Br. J. Cancer 1993, 67,
1163-1170.
(11) Denny, W. A.; Wilson, W. R. Tirapazamine: a bioreductive
anticancer drug that exploits tumour hypoxia. Expert Opin.
Invest. Drugs 2000, 9, 2889-2901.
(12) Baker, M. A.; Zeman, E. M.; Hirst, V. K.; Brown, J. M.
Metabolism of SR 4233 by Chinese hamster ovary cells: basis
of selective hypoxic cytotoxicity. Cancer Res. 1988, 48, 5947-
5952.
(13) Siim, B. G.; van Zijl, P. L.; Brown, J. M. Tirapazamine-induced
DNA damage measured using the comet assay correlates with
cytotoxicity towards hypoxic tumour cells in vitro. Br. J. Cancer
1996, 73, 952-960.
(14) Durand, R. E.; Olive, P. L. Evaluation of bioreductive drugs in
multicell spheroids. Int. J. Radiat. Oncol., Biol., Phys. 1992, 22,
689-692.
(31) Cummings, J.; Boyd, G.; Macpherson, J. S.; Wolf, H.; Smith, G.;
Smyth, J. F.; Jodrell, D. I. Factors influencing the cellular
accumulation of SN-38 and camptothecin. Cancer Chemother.
Pharmacol. 2002, 49, 194-200.
(32) Schmittmann, G.; Rohr, U. D. Comparison of the permeability
surface product (PS) of the blood capillary wall in skeletal muscle
tissue of various species and in vitro porous membranes using
hydrophilic drugs. J. Pharm. Sci. 2000, 89, 115-127.
(33) Artursson, P. Epithelial transport of drugs in cell culture. I: A
model for studying the passive diffusion of drugs over intestinal
absorptive (Caco-2) cells. J. Pharm. Sci. 1990, 79, 476-482.
(34) Adson, A.; Burton, P. S.; Raub, T. J.; Barsuhn, C. L.; Audus, K.
L.; Ho, N. F. Passive diffusion of weak organic electrolytes across
Caco-2 cell monolayers: uncoupling the contributions of hydro-
dynamic, transcellular, and paracellular barriers. J. Pharm. Sci.
1995, 84, 1197-1204.
(35) Dearden, J. C. Molecular structure and drug transport. In
Comprehensive Medicinal Chemistry: The Rational Design,
Mechanistic Study & Therapeutic Application of Chemical
Compounds, 1st ed.; Hansch, C., Sammes, P. G., Taylor, J. B.,
Eds.; Pergamon Press: Oxford and New York, 1990; pp 375-
411.
(36) Mastro, A. M.; Babich, M. A.; Taylor, W. D.; Keith, A. D.
Diffusion of a small molecule in the cytoplasm of mammalian
cells. Proc. Natl. Acad. Sci. U.S.A. 1984, 81, 3414-3418.
(37) Hicks, K. O.; Siim, B. G.; Pruijn, F. B.; Wilson, W. R. Oxygen
dependence of the metabolic activation and cytotoxicity of
tirapazamine: implications for extravascular transport and
activity in tumors. Radiat. Res. 2004, 161, 656-666.
(38) Zeman, E. M.; Baker, M. A.; Lemmon, M. J.; Pearson, C. I.;
Adams, J. A.; Brown, J. M.; Lee, W. W.; Tracy, M. Structure-
activity relationships for benzotriazine di-N-oxides. Int. J.
Radiat. Oncol., Biol., Phys. 1989, 16, 977-981.
(39) Kelson, A. B.; McNamara, J. P.; Pandey, A.; Ryan, K. J.; Dorie,
M. J.; McAfee, P. A.; Menke, D. R.; Brown, J. M.; Tracy, M. 1,2,4-
Benzotriazine 1,4-dioxides. An important class of hypoxic cyto-
toxins with antitumor activity. Anti-Cancer Drug Des. 1998, 13,
575-592.
(15) Hicks, K. O.; Fleming, Y.; Siim, B. G.; Koch, C. J.; Wilson, W.
R. Extravascular diffusion of tirapazamine: effect of metabolic
consumption assessed using the multicellular layer model. Int.
J. Radiat. Oncol., Biol., Phys. 1998, 42, 641-649.
(16) Kyle, A. H.; Minchinton, A. I. Measurement of delivery and
metabolism of tirapazamine to tumour tissue using the multi-
layered cell culture model. Cancer Chemother. Pharmacol. 1999,
43, 213-220.
(17) Baguley, B. C.; Hicks, K. O.; Wilson, W. R. Tumour cell cultures
in drug development. In Anticancer Drug Development; Baguley,
B. C., Kerr, D. J., Eds.; Academic Press: San Diego, CA, 2002;
pp 269-284.
(18) Hicks, K. O.; Pruijn, F. B.; Sturman, J. R.; Denny, W. A.; Wilson,
W. R. Multicellular resistance to tirapazamine is due to re-
stricted extravascular transport: a pharmacokinetic/pharma-
codynamic study in HT29 multicellular layer cultures. Cancer
Res. 2003, 63, 5970-5977.
(40) Goodwin, J. T.; Conradi, R. A.; Ho, N. F.; Burton, P. S.
Physicochemical determinants of passive membrane perme-
ability: role of solute hydrogen-bonding potential and volume.
J. Med. Chem. 2001, 44, 3721-3729.

Extravascular Transport of Tirapazamine Analogues
Journal of Medicinal Chemistry, 2005, Vol. 48, No. 4 1087
(41) Stenberg, P.; Luthman, K.; Artursson, P. Prediction of membrane
permeability to peptides from calculated dynamic molecular
surface properties. Pharm. Res. 1999, 16, 205-212.
(42) Brown, J. M.; Giaccia, A. J. The unique physiology of solid
tumors: opportunities (and problems) for cancer therapy. Cancer
Res. 1998, 58, 1408-1416.
D.; Arcenas, A.; Vescio, R.; Viallet, J.; Treat, J. Tirapazamine
plus cisplatin versus cisplatin in advanced non-small-cell lung
cancer: A report of the international CATAPULT I study group.
Cisplatin and tirapazamine in subjects with advanced previously
untreated non-small-cell lung tumors. J. Clin. Oncol. 2000, 18,
1351-1359.
(43) Lankelma, J. Tissue transport of anti-cancer drugs. Curr. Pharm.
Des. 2002, 8, 1987-1993.
(50) Bedikian, A. Y.; Legha, S. S.; Eton, O.; Buzaid, A. C.; Papa-
dopoulos, N.; Plager, C.; McIntyre, S.; Viallet, J. Phase II trial
of escalated dose of tirapazamine combined with cisplatin in
advanced malignant melanoma. Anti-Cancer Drugs 1999, 10,
735-739.
(51) Graham, M. A.; Senan, S.; Robin, H., Jr.; Eckhardt, N.; Lendrem,
D.; Hincks, J.; Greenslade, D.; Rampling, R.; Kaye, S. B.; von
Roemeling, R.; Workman, P. Pharmacokinetics of the hypoxic
cell cytotoxic agent tirapazamine and its major bioreductive
metabolites in mice and humans: retrospective analysis of a
pharmacokinetically guided dose-escalation strategy in a phase
I trial. Cancer Chemother. Pharmacol. 1997, 40, 1-10.
(52) Johnson, C. A.; Kilpatrick, D.; von Roemeling, R.; Langer, C.;
Graham, M. A.; Greenslade, D.; Kennedy, G.; Keenan, E.;
O’Dwyer, P. J. Phase I trial of tirapazamine in combination with
cisplatin in a single dose every 3 weeks in patients with solid
tumors. J. Clin. Oncol. 1997, 15, 773-780.
(44) Tannock, I. F. Tumor physiology and drug resistance. Cancer
Metastasis Rev. 2001, 20, 123-132.
(45) Grote, J.; Su¨sskind, R.; Vaupel, P. Oxygen diffusivity in tumor
tissue (DS-carcinosarcoma) under temperature conditions within
the range of 20-40 °C. Pfluegers Arch. 1977, 372, 37-42.
(46) Dewhirst, M. W.; Secomb, T. W.; Ong, E. T.; Hsu, R.; Gross, J.
F. Determination of local oxygen consumption rates in tumors.
Cancer Res. 1994, 54, 3333-3336.
(47) Dewhirst, M. W.; Ong, E. T.; Klitzman, B.; Secomb, T. W.;
Vinuya, R. Z.; Dodge, R.; Brizel, D.; Gross, J. F. Perivascular
oxygen tensions in a transplantable mammary tumor growing
in a dorsal flap window chamber. Radiat. Res. 1992, 130, 171-
182.
(48) Hoff, P. M.; Saad, E. D.; Ravandi-Kashani, F.; Czerny, E.;
Pazdur, R. Phase I trial of i.v. administered tirapazamine plus
cyclophosphamide. Anti-Cancer Drugs 2001, 12, 499-503.
(49) von Pawel, J.; von Roemeling, R.; Gatzemeier, U.; Boyer, M.;
Elisson, L. O.; Clark, P.; Talbot, D.; Rey, A.; Butler, T. W.; Hirsh,
V.; Olver, I.; Bergman, B.; Ayoub, J.; Richardson, G.; Dunlop,
JM049549P